Search

Multi-agent systems - systems with multiple independent AI agents working together to achieve a common goal - are becoming increasingly prevalent in daily life. Drawing inspiration from the phenomenon of human group social influence, we investigate whether a group of AI agents can create social pressure on users to agree with them, potentially changing their stance on a topic. We conducted a study in which participants discussed social issues with either a single or multiple AI agents, and where the agents either agreed or disagreed with the user's stance on the topic. We found that conversing with multiple agents (holding conversation content constant) increased the social pressure felt by participants, and caused a greater shift in opinion towards the agents' stances on each topic. Our study shows the potential advantages of multi-agent systems over single-agent platforms in causing opinion change. We discuss design implications for possible multi-agent systems that promote social good, as well as the potential for malicious actors to use these systems to manipulate public opinion.

Current RLHF frameworks for aligning large language models (LLMs) typically assume a fixed prompt distribution, which is sub-optimal and limits the scalability of alignment and generalizability of models. To address this, we introduce a general open-ended RLHF framework that casts alignment as an asymmetric game between two players: (i) a creator that generates increasingly informative prompt distributions using the reward model, and (ii) a solver that learns to produce more preferred responses on prompts produced by the creator. This framework of Evolving Alignment via Asymmetric Self-Play (eva), results in a simple and efficient approach that can utilize any existing RLHF algorithm for scalable alignment. eva outperforms state-of-the-art methods on widely-used benchmarks, without the need of any additional human crafted prompts. Specifically, eva improves the win rate of Gemma-2-9B-it on Arena-Hard from 51.6% to 60.1% with DPO, from 55.7% to 58.9% with SPPO, from 52.3% to 60.7% with SimPO, and from 54.8% to 60.3% with ORPO, surpassing its 27B version and matching claude-3-opus. This improvement is persistent even when new human crafted prompts are introduced. Finally, we show eva is effective and robust under various ablation settings.

Quantum secret sharing (QSS) plays a pivotal role in multiparty quantum communication, ensuring the secure distribution of private information among multiple parties. However, the security of QSS schemes can be compromised by attacks exploiting imperfections in measurement devices. Here, we propose a reconfigurable approach to implement QSS based on measurement-device-independent (MDI) principles, utilizing linear two-photon Bell state analysis.By employing single-qubit conjugate operations for encoding private information, our approach offers reconfigurability, allowing for the inclusion of additional quantum network nodes without sacrificing efficiency. Furthermore, we demonstrate the robust security of our MDI-QSS scheme against inter-eavesdropping by dishonest participants and establish lower bounds for secure communication among three legitimate parties. This work presents a flexible configuration for implementing multiparty secure quantum communication with imperfect measurement devices and represents a significant advancement in the development of secure quantum communication technologies., Comment: 9 pages, 3 figures

Heterogeneous multi-robot systems (HMRS) have emerged as a powerful approach for tackling complex tasks that single robots cannot manage alone. Current large-language-model-based multi-agent systems (LLM-based MAS) have shown success in areas like software development and operating systems, but applying these systems to robot control presents unique challenges. In particular, the capabilities of each agent in a multi-robot system are inherently tied to the physical composition of the robots, rather than predefined roles. To address this issue, we introduce a novel multi-agent framework designed to enable effective collaboration among heterogeneous robots with varying embodiments and capabilities, along with a new benchmark named Habitat-MAS. One of our key designs is $\textit{Robot Resume}$: Instead of adopting human-designed role play, we propose a self-prompted approach, where agents comprehend robot URDF files and call robot kinematics tools to generate descriptions of their physics capabilities to guide their behavior in task planning and action execution. The Habitat-MAS benchmark is designed to assess how a multi-agent framework handles tasks that require embodiment-aware reasoning, which includes 1) manipulation, 2) perception, 3) navigation, and 4) comprehensive multi-floor object rearrangement. The experimental results indicate that the robot's resume and the hierarchical design of our multi-agent system are essential for the effective operation of the heterogeneous multi-robot system within this intricate problem context., Comment: 10 pages of main content, 3 pages of references, 5 pages of appendix, 7 figures in total

Query Autocomplete (QAC) is a critical feature in modern search engines, facilitating user interaction by predicting search queries based on input prefixes. Despite its widespread adoption, the absence of large-scale, realistic datasets has hindered advancements in QAC system development. This paper addresses this gap by introducing AmazonQAC, a new QAC dataset sourced from Amazon Search logs, comprising 395M samples. The dataset includes actual sequences of user-typed prefixes leading to final search terms, as well as session IDs and timestamps that support modeling the context-dependent aspects of QAC. We assess Prefix Trees, semantic retrieval, and Large Language Models (LLMs) with and without finetuning. We find that finetuned LLMs perform best, particularly when incorporating contextual information. However, even our best system achieves only half of what we calculate is theoretically possible on our test data, which implies QAC is a challenging problem that is far from solved with existing systems. This contribution aims to stimulate further research on QAC systems to better serve user needs in diverse environments. We open-source this data on Hugging Face at https://huggingface.co/datasets/amazon/AmazonQAC., Comment: EMNLP 2024

Automated Machine Learning (AutoML) approaches encompass traditional methods that optimize fixed pipelines for model selection and ensembling, as well as newer LLM-based frameworks that autonomously build pipelines. While LLM-based agents have shown promise in automating machine learning tasks, they often generate low-diversity and suboptimal code, even after multiple iterations. To overcome these limitations, we introduce Tree-Search Enhanced LLM Agents (SELA), an innovative agent-based system that leverages Monte Carlo Tree Search (MCTS) to optimize the AutoML process. By representing pipeline configurations as trees, our framework enables agents to conduct experiments intelligently and iteratively refine their strategies, facilitating a more effective exploration of the machine learning solution space. This novel approach allows SELA to discover optimal pathways based on experimental feedback, improving the overall quality of the solutions. In an extensive evaluation across 20 machine learning datasets, we compare the performance of traditional and agent-based AutoML methods, demonstrating that SELA achieves a win rate of 65% to 80% against each baseline across all datasets. These results underscore the significant potential of agent-based strategies in AutoML, offering a fresh perspective on tackling complex machine learning challenges., Comment: The code is available at https://github.com/geekan/MetaGPT

Probabilistic Bit (P-Bit) device serves as the core hardware for implementing Ising computation. However, the severe intrinsic variations of stochastic P-Bit devices hinder the large-scale expansion of the P-Bit array, significantly limiting the practical usage of Ising computation. In this work, a behavioral model which attributes P-Bit variations to two parameters {\alpha} and {\Delta}V is proposed. Then the weight compensation method is introduced, which can mitigate {\alpha} and {\Delta}V of P-Bits device variations by rederiving the weight matrix, enabling them to compute as ideal identical PBits without the need for weights retraining. Accurately extracting the {\alpha} and {\Delta}V simultaneously from a large P-Bit array which is prerequisite for the weight compensation method is a crucial and challenging task. To solve this obstacle, we present the novel automatic variation extraction algorithm which can extract device variations of each P-Bit in a large array based on Boltzmann machine learning. In order for the accurate extraction of variations from an extendable P-Bit array, an Ising Hamiltonian based on 3D ferromagnetic model is constructed, achieving precise and scalable array variation extraction. The proposed Automatic Extraction and Compensation algorithm is utilized to solve both 16-city traveling salesman problem(TSP) and 21-bit integer factorization on a large P-Bit array with variation, demonstrating its accuracy, transferability, and scalability., Comment: 15 pages, 17 figures

KV cache has become a de facto technique for the inference of large language models (LLMs), where tensors of shape (layer number, head number, sequence length, feature dimension) are introduced to cache historical information for self-attention. As the size of the model and data grows, the KV cache can quickly become a bottleneck within the system in both storage and memory transfer. To address this, prior studies usually focus on the first three axes of the cache tensors for compression. This paper supplements them, focusing on the feature dimension axis, by utilizing low-rank projection matrices to transform the cache features into spaces with reduced dimensions. We begin by investigating the canonical orthogonal projection method for data compression through principal component analysis (PCA). We observe the issue with PCA projection where significant performance degradation is observed at low compression rates. To bridge the gap, we propose to directly tune the orthogonal projection matrices with a distillation objective using an elaborate Matryoshka training strategy. After training, we adaptively search for the optimal compression rates for various layers and heads given varying compression budgets. Compared to previous works, our method can easily embrace pre-trained LLMs and hold a smooth tradeoff between performance and compression rate. We empirically witness the high data efficiency of our training procedure and find that our method can sustain over 90% performance with an average KV cache compression rate of 60% (and up to 75% in certain extreme scenarios) for popular LLMs like LLaMA2-7B-base and Mistral-7B-v0.3-base.

The KV-Cache technique has become the standard for the inference of large language models (LLMs). It caches states of self-attention to avoid recomputation. Yet, it is widely criticized that KV-Cache can become a bottleneck of the LLM inference system, especially when confronted with ultra-large models and long-context queries. A natural remedy is to discard the KV-Cache for less important tokens, with StreamingLLM as an example, but the used static eviction strategies cannot flexibly adapt to varying contexts. Remedies like H2O leverage accumulative attention scores to perform dynamic eviction but suffer from the attention bias issue in capturing contextual information. This paper bridges this gap by devising a parameterized KV-Cache eviction mechanism, dubbed as Attention-Gate, which accepts the whole context as input and yields eviction flags for each token to realize in-context eviction. The subsequent self-attention module proceeds according to the flags and only the KV states for the remaining tokens need to be cached. The Attention-Gates can vary among different heads and layers and be trivially plugged into pre-trained LLMs, tuned by cost-effective continual pre-training or supervised fine-tuning objectives to acquire what to discard. The computational and memory overhead introduced by Attention-Gates is minimal. Our method is validated across multiple tasks, demonstrating both efficiency and adaptability. After a highly efficient continual pre-training, it achieves higher average accuracy and evicts more tokens compared to traditional training-free methods. In supervised fine-tuning, it not only evicts many tokens but also outperforms LoRA-finetuned LLMs on some datasets, such as RTE, where it improves accuracy by 13.9% while evicting 62.8% of tokens, showing that effective eviction of redundant tokens can even enhance performance.

The standard model of Boolean function property testing is not well suited for testing $\textit{sparse}$ functions which have few satisfying assignments, since every such function is close (in the usual Hamming distance metric) to the constant-0 function. In this work we propose and investigate a new model for property testing of Boolean functions, called $\textit{relative-error testing}$, which provides a natural framework for testing sparse functions. This new model defines the distance between two functions $f, g: \{0,1\}^n \to \{0,1\}$ to be $$\textsf{reldist}(f,g) := { \frac{|f^{-1}(1) \triangle g^{-1}(1)|} {|f^{-1}(1)|}}.$$ This is a more demanding distance measure than the usual Hamming distance ${ {|f^{-1}(1) \triangle g^{-1}(1)|}/{2^n}}$ when $|f^{-1}(1)| \ll 2^n$; to compensate for this, algorithms in the new model have access both to a black-box oracle for the function $f$ being tested and to a source of independent uniform satisfying assignments of $f$. In this paper we first give a few general results about the relative-error testing model; then, as our main technical contribution, we give a detailed study of algorithms and lower bounds for relative-error testing of $\textit{monotone}$ Boolean functions. We give upper and lower bounds which are parameterized by $N=|f^{-1}(1)|$, the sparsity of the function $f$ being tested. Our results show that there are interesting differences between relative-error monotonicity testing of sparse Boolean functions, and monotonicity testing in the standard model. These results motivate further study of the testability of Boolean function properties in the relative-error model.

We introduce PLaMo-100B, a large-scale language model designed for Japanese proficiency. The model was trained from scratch using 2 trillion tokens, with architecture such as QK Normalization and Z-Loss to ensure training stability during the training process. Post-training techniques, including Supervised Fine-Tuning and Direct Preference Optimization, were applied to refine the model's performance. Benchmark evaluations suggest that PLaMo-100B performs well, particularly in Japanese-specific tasks, achieving results that are competitive with frontier models like GPT-4. The base model is available at https://huggingface.co/pfnet/plamo-100b.

Group re-identification (re-ID) aims to match groups with the same people under different cameras, mainly involves the challenges of group members and layout changes well. Most existing methods usually use the k-nearest neighbor algorithm to update node features to consider changes in group membership, but these methods cannot solve the problem of group layout changes. To this end, we propose a novel vision transformer based random walk framework for group re-ID. Specifically, we design a vision transformer based on a monocular depth estimation algorithm to construct a graph through the average depth value of pedestrian features to fully consider the impact of camera distance on group members relationships. In addition, we propose a random walk module to reconstruct the graph by calculating affinity scores between target and gallery images to remove pedestrians who do not belong to the current group. Experimental results show that our framework is superior to most methods., Comment: 6 pages

Growing evidence suggests that social determinants of health (SDoH), a set of nonmedical factors, affect individuals' risks of developing Alzheimer's disease (AD) and related dementias. Nevertheless, the etiological mechanisms underlying such relationships remain largely unclear, mainly due to difficulties in collecting relevant information. This study presents a novel, automated framework that leverages recent advancements of large language model (LLM) and natural language processing techniques to mine SDoH knowledge from extensive literature and integrate it with AD-related biological entities extracted from the general-purpose knowledge graph PrimeKG. Utilizing graph neural networks, we performed link prediction tasks to evaluate the resultant SDoH-augmented knowledge graph. Our framework shows promise for enhancing knowledge discovery in AD and can be generalized to other SDoH-related research areas, offering a new tool for exploring the impact of social determinants on health outcomes. Our code is available at: https://github.com/hwq0726/SDoHenPKG

Creating data videos that effectively narrate stories with animated visuals requires substantial effort and expertise. A promising research trend is leveraging the easy-to-use natural language (NL) interaction to automatically synthesize data video components from narrative content like text narrations, or NL commands that specify user-required designs. Nevertheless, previous research has overlooked the integration of narrative content and specific design authoring commands, leading to generated results that lack customization or fail to seamlessly fit into the narrative context. To address these issues, we introduce a novel paradigm for creating data videos, which seamlessly integrates users' authoring and narrative intents in a unified format called annotated narration, allowing users to incorporate NL commands for design authoring as inline annotations within the narration text. Informed by a formative study on users' preference for annotated narration, we develop a prototype system named Data Playwright that embodies this paradigm for effective creation of data videos. Within Data Playwright, users can write annotated narration based on uploaded visualizations. The system's interpreter automatically understands users' inputs and synthesizes data videos with narration-animation interplay, powered by large language models. Finally, users can preview and fine-tune the video. A user study demonstrated that participants can effectively create data videos with Data Playwright by effortlessly articulating their desired outcomes through annotated narration., Comment: 11 pages, 7 figures, accepted by IEEE TVCG

Affine frequency division multiplexing (AFDM), tailored as a superior multicarrier technique utilizing chirp signals for high-mobility communications, is envisioned as a promising candidate for the sixth-generation (6G) wireless network. AFDM is based on the discrete affine Fourier transform (DAFT) with two adjustable parameters of the chirp signals, termed as the pre-chirp and post-chirp parameters, respectively. We show that the pre-chirp counterpart can be flexibly manipulated for additional degree-of-freedom (DoF). Therefore, this paper proposes a novel AFDM scheme with the pre-chirp index modulation (PIM) philosophy (AFDM-PIM), which can implicitly convey extra information bits through dynamic pre-chirp parameter assignment, thus enhancing both spectral and energy efficiency. Specifically, we first demonstrate that the subcarrier orthogonality is still maintained by applying distinct pre-chirp parameters to various subcarriers in the AFDM modulation process. Inspired by this property, each AFDM subcarrier is constituted with a unique pre-chirp signal according to the incoming bits. By such arrangement, extra binary bits can be embedded into the index patterns of pre-chirp parameter assignment without additional energy consumption. For performance analysis, we derive the asymptotically tight upper bounds on the average bit error rates (BERs) of the proposed schemes with maximum-likelihood (ML) detection, and validate that the proposed AFDM-PIM can achieve the optimal diversity order under doubly dispersive channels. Based on the derivations, we further propose an optimal pre-chirp alphabet design to enhance the BER performance via intelligent optimization algorithms. Simulations demonstrate that the proposed AFDM-PIM outperforms the classical benchmarks under doubly dispersive channel.

KIC 10685175 is a roAp star whose polar magnetic field is predicted to be 6 kG through a non-adiabatic axisymmetric pulsation theoretical model. In this work, we aim to measure the magnetic field strength of KIC 10685175 using high-resolution spectropolarimetric observations, and compare it with the one predicted by the theoretical model. From the study of two high-resolution unpolarized spectra, we obtained [$T_{\rm eff}$, $\log g$, [Fe/H], [$\alpha$/Fe], $V_{mic}$]=[8250 $\pm$ 200\,K, 4.4 $\pm$ 0.1, -0.4 $\pm$ 0.2, 0.16 $\pm$ 0.1, 1.73 $\pm$ 0.2\,km~s$^{-1}$]. Although the Fe absorption lines appear relatively weak in comparison to typical Ap stars with similar $T_{\rm eff}$, the lines belonging to rare earth elements (Eu and Nd) are stronger than that in chemically normal stars, indicating the peculiar nature of KIC~10685175. The mean longitudinal magnetic field $\langle B_\ell \rangle=-226\pm39$\,G has been measured in the polarized spectrum, but magnetically split lines were not detected. No significant line profile variability was evident in our spectra. Also the longitudinal magnetic field strengths measured using line masks constructed for different elements have been rather similar. Due to a poor rotation phase coverage of our data, additional spectroscopic and polarimetric observations are needed to allow us to conclude on the inhomogeneous element distribution over the stellar surface. The estimated polar magnetic field is $4.8 \pm 0.8$\,kG, which is consistent with the predicted polar magnetic field strength of about 6\,kG within 3$\sigma$. This work therefore provides support for the pulsation theoretical model.

Utilizing Self-Supervised Learning (SSL) models for Speech Emotion Recognition (SER) has proven effective, yet limited research has explored cross-lingual scenarios. This study presents a comparative analysis between human performance and SSL models, beginning with a layer-wise analysis and an exploration of parameter-efficient fine-tuning strategies in monolingual, cross-lingual, and transfer learning contexts. We further compare the SER ability of models and humans at both utterance- and segment-levels. Additionally, we investigate the impact of dialect on cross-lingual SER through human evaluation. Our findings reveal that models, with appropriate knowledge transfer, can adapt to the target language and achieve performance comparable to native speakers. We also demonstrate the significant effect of dialect on SER for individuals without prior linguistic and paralinguistic background. Moreover, both humans and models exhibit distinct behaviors across different emotions. These results offer new insights into the cross-lingual SER capabilities of SSL models, underscoring both their similarities to and differences from human emotion perception.

Manipulating fragile objects in environments such as homes and factories requires stable and gentle grasping along with precise and safe placement. Compared to traditional rigid grippers, the use of soft grippers reduces the control complexity and the risk of damaging objects. However, it is challenging to integrate camera-based optical tactile sensing into a soft gripper without compromising the flexibility and adaptability of the fingers, while also ensuring that the precision of tactile perception remains unaffected by passive deformations of the soft structure during object contact. In this paper, we demonstrate a modular soft two-fingered gripper with a 3D-printed optical tactile sensor (the TacTip) integrated in the palm. We propose a soft-grasping strategy that includes three functions: light contact detection, grasp pose adjustment and loss-of-contact detection, so that objects of different shapes and sizes can be grasped stably and placed precisely, which we test with both artificial and household objects. By sequentially implementing these three functions, the grasp success rate progressively improves from 45% without any functions, to 59% with light contact detection, 90% with grasp pose adjustment, and 97% with loss-of-contact detection, achieving a sub-millimeter placement precision. Overall, this work demonstrates the feasibility and utility of integrating optical tactile sensors into the palm of a soft gripper, of applicability to various types of soft manipulators. The proposed grasping strategy has potential applications in areas such as fragile product processing and home assistance.

Safety is one of the most crucial challenges of autonomous driving vehicles, and one solution to guarantee safety is to employ an additional control revision module after the planning backbone. Control Barrier Function (CBF) has been widely used because of its strong mathematical foundation on safety. However, the incompatibility with heterogeneous perception data and incomplete consideration of traffic scene elements make existing systems hard to be applied in dynamic and complex real-world scenarios. In this study, we introduce a generalized control revision method for autonomous driving safety, which adopts both vectorized perception and occupancy grid map as inputs and comprehensively models multiple types of traffic scene constraints based on a new proposed barrier function. Traffic elements are integrated into one unified framework, decoupled from specific scenario settings or rules. Experiments on CARLA, SUMO, and OnSite simulator prove that the proposed algorithm could realize safe control revision under complicated scenes, adapting to various planning backbones, road topologies, and risk types. Physical platform validation also verifies the real-world application feasibility.

Grain boundaries have extensive influence on the performance of crystal materials. However, the atomic-scale structure and its relation with local and crystallographic symmetries remain elusive in low-symmetry crystals. Herein, we find that the local pseudo-mirror-symmetric atomic layer is the common physical origin of a series of highly coherent grain boundaries in the low-symmetry $\beta$-Ga2O3 crystal. These include the (100) twin boundary and an emerging series of $(h-1'0'2)/(h+1'0'\bar{2})$ coherent asymmetric grain boundaries (CAGBs). Owing to the local pseudo-mirror symmetry and the special geometric relation of the $\beta$-Ga2O3 conventional cell, these CAGBs place 80% of the boundary atoms in pseudo-coincident sites, exhibiting high coherence under the coincident-site lattice model. With a combination of density functional theory calculations, Czochralski growth experiment, and atomic-scale characterizations, the structure and stability of the $(002)/(20\bar{2})$-A CAGB are confirmed, with a boundary energy density as low as 0.36 J/m2. This CAGB is responsible for the spontaneous formation of a twinned defect facet at the surface steps during the epitaxy growth of $\beta$-Ga2O3, warranting a substrate orientation selection rule for $\beta$-Ga2O3. Through this study, we provide insights into the grain boundary physics in the low-symmetry $\beta$-Ga2O3 crystal while emphasizing the importance of the local pseudo-symmetries in the low-symmetry crystals.

In this paper, we propose an emerging wavenumber-domain precoding scheme to break the limitations of rank-1 channels that merely supports single-stream transmission, enabling simultaneous transmission of multiple data streams. The proposed wavenumber-domain precoding scheme also breaks the Rayleigh distance demarcation, regardless of the far-field and near-field contexts. Specifically, by characterizing the channel response as the superposition of a series of Fourier harmonics specified by different wavenumbers, the degree of freedom (DoF) is dependent on the cardinality of the wavenumber support, based on which the extra DoF is presented. This representation is applicable for both far-field and near-field. Different wavenumber atoms, determined within this support, constitute the codebook for MIMO precoding, in which each atom allows for the transmission of a data stream. Then, to maximize the capacity, it is required to select the wavenumbers associated with the optimal transmission direction, and optimize its power allocation. Finally, our simulation results demonstrate the significant superiority in comparison to the conventional spatial division schemes, with the potential of approaching the theoretical performance upper bound achieved by singular value decomposition (SVD)., Comment: This paper has been accepted in 2024 IEEE Globecom Workshop

Reward models (RMs) play a pivotal role in aligning large language models (LLMs) with human preferences. However, traditional RM training, which relies on response pairs tied to specific prompts, struggles to disentangle prompt-driven preferences from prompt-independent artifacts, such as response length and format. In this work, we expose a fundamental limitation of current RM training methods, where RMs fail to effectively distinguish between contextual signals and irrelevant artifacts when determining preferences. To address this, we introduce a causal framework that learns preferences independent of these artifacts and propose a novel data augmentation technique designed to eliminate them. Extensive experiments show that our approach successfully filters out undesirable artifacts, yielding a more robust reward model (RRM). Our RRM improves the performance of a pairwise reward model trained on Gemma-2-9b-it, on RewardBench, increasing accuracy from 80.61% to 84.15%. Additionally, we train two DPO policies using both the RM and RRM, demonstrating that the RRM significantly enhances DPO-aligned policies, improving MT-Bench scores from 7.27 to 8.31 and length-controlled win-rates in AlpacaEval-2 from 33.46% to 52.49%.

The machine learning community is increasingly recognizing the importance of fostering trust and safety in modern generative AI (GenAI) models. We posit machine unlearning (MU) as a crucial foundation for developing safe, secure, and trustworthy GenAI models. Traditional MU methods often rely on stringent assumptions and require access to real data. This paper introduces Score Forgetting Distillation (SFD), an innovative MU approach that promotes the forgetting of undesirable information in diffusion models by aligning the conditional scores of "unsafe" classes or concepts with those of "safe" ones. To eliminate the need for real data, our SFD framework incorporates a score-based MU loss into the score distillation objective of a pretrained diffusion model. This serves as a regularization term that preserves desired generation capabilities while enabling the production of synthetic data through a one-step generator. Our experiments on pretrained label-conditional and text-to-image diffusion models demonstrate that our method effectively accelerates the forgetting of target classes or concepts during generation, while preserving the quality of other classes or concepts. This unlearned and distilled diffusion not only pioneers a novel concept in MU but also accelerates the generation speed of diffusion models. Our experiments and studies on a range of diffusion models and datasets confirm that our approach is generalizable, effective, and advantageous for MU in diffusion models. (Warning: This paper contains sexually explicit imagery, discussions of pornography, racially-charged terminology, and other content that some readers may find disturbing, distressing, and/or offensive.)

We study the problem of generating intermediate images from image pairs with large motion while maintaining semantic consistency. Due to the large motion, the intermediate semantic information may be absent in input images. Existing methods either limit to small motion or focus on topologically similar objects, leading to artifacts and inconsistency in the interpolation results. To overcome this challenge, we delve into pre-trained image diffusion models for their capabilities in semantic cognition and representations, ensuring consistent expression of the absent intermediate semantic representations with the input. To this end, we propose DreamMover, a novel image interpolation framework with three main components: 1) A natural flow estimator based on the diffusion model that can implicitly reason about the semantic correspondence between two images. 2) To avoid the loss of detailed information during fusion, our key insight is to fuse information in two parts, high-level space and low-level space. 3) To enhance the consistency between the generated images and input, we propose the self-attention concatenation and replacement approach. Lastly, we present a challenging benchmark dataset InterpBench to evaluate the semantic consistency of generated results. Extensive experiments demonstrate the effectiveness of our method. Our project is available at https://dreamm0ver.github.io ., Comment: ECCV 2024

Managing fluid balance in dialysis patients is crucial, as improper management can lead to severe complications. In this paper, we propose a multimodal approach that integrates visual features from lung ultrasound images with clinical data to enhance the prediction of excess body fluid. Our framework employs independent encoders to extract features for each modality and combines them through a cross-domain attention mechanism to capture complementary information. By framing the prediction as a classification task, the model achieves significantly better performance than regression. The results demonstrate that multimodal models consistently outperform single-modality models, particularly when attention mechanisms prioritize tabular data. Pseudo-sample generation further contributes to mitigating the imbalanced classification problem, achieving the highest accuracy of 88.31%. This study underscores the effectiveness of multimodal learning for fluid overload management in dialysis patients, offering valuable insights for improved clinical outcomes., Comment: In the experiment, for the classification tasks, the network was informed with ground truth during training, significantly improving the performance. This makes the results invalid. Therefore, corrections and more validations are needed to evaluate the performance of the method

In modern agriculture, precise monitoring of plants and fruits is crucial for tasks such as high-throughput phenotyping and automated harvesting. This paper addresses the challenge of reconstructing accurate 3D shapes of fruits from partial views, which is common in agricultural settings. We introduce CF-PRNet, a coarse-to-fine prototype refining network, leverages high-resolution 3D data during the training phase but requires only a single RGB-D image for real-time inference. Our approach begins by extracting the incomplete point cloud data that constructed from a partial view of a fruit with a series of convolutional blocks. The extracted features inform the generation of scaling vectors that refine two sequentially constructed 3D mesh prototypes - one coarse and one fine-grained. This progressive refinement facilitates the detailed completion of the final point clouds, achieving detailed and accurate reconstructions. CF-PRNet demonstrates excellent performance metrics with a Chamfer Distance of 3.78, an F1 Score of 66.76%, a Precision of 56.56%, and a Recall of 85.31%, and win the first place in the Shape Completion and Reconstruction of Sweet Peppers Challenge., Comment: Technical Report of the 1st place solution to CVPPA@ECCV2024: Shape Completion and Reconstruction of Sweet Peppers Challenge

Sparse Regression Codes (SPARCs) are capacity-achieving codes introduced for communication over the Additive White Gaussian Noise (AWGN) channels and were later extended to general memoryless channels. In particular it was shown via threshold saturation that Spatially Coupled Sparse Regression Codes (SC-SPARCs) are capacity-achieving over general memoryless channels when using an Approximate Message Passing decoder (AMP). This paper, for the first time rigorously, analyzes the non-asymptotic performance of the Generalized Approximate Message Passing (GAMP) decoder of SC-SPARCs over memoryless channels, and proves exponential decaying error probability with respect to the code length.

Plant diseases pose significant threats to agriculture. It necessitates proper diagnosis and effective treatment to safeguard crop yields. To automate the diagnosis process, image segmentation is usually adopted for precisely identifying diseased regions, thereby advancing precision agriculture. Developing robust image segmentation models for plant diseases demands high-quality annotations across numerous images. However, existing plant disease datasets typically lack segmentation labels and are often confined to controlled laboratory settings, which do not adequately reflect the complexity of natural environments. Motivated by this fact, we established PlantSeg, a large-scale segmentation dataset for plant diseases. PlantSeg distinguishes itself from existing datasets in three key aspects. (1) Annotation type: Unlike the majority of existing datasets that only contain class labels or bounding boxes, each image in PlantSeg includes detailed and high-quality segmentation masks, associated with plant types and disease names. (2) Image source: Unlike typical datasets that contain images from laboratory settings, PlantSeg primarily comprises in-the-wild plant disease images. This choice enhances the practical applicability, as the trained models can be applied for integrated disease management. (3) Scale: PlantSeg is extensive, featuring 11,400 images with disease segmentation masks and an additional 8,000 healthy plant images categorized by plant type. Extensive technical experiments validate the high quality of PlantSeg's annotations. This dataset not only allows researchers to evaluate their image classification methods but also provides a critical foundation for developing and benchmarking advanced plant disease segmentation algorithms.

This paper proposes a transmit beamforming strategy for the integrated sensing and communication (ISAC) systems enabled by the novel stacked intelligent metasurface (SIM) architecture, where the base station (BS) simultaneously performs downlink communication and radar target detection via different beams. To ensure superior dual-function performance simultaneously, we design the multi-layer cascading beamformer by maximizing the sum rate of the users while optimally shaping the normalized beam pattern for detection. A dual-normalized differential gradient descent (D3) algorithm is further proposed to solve the resulting non-convex multi-objective problem (MOP), where gradient differences and dual normalization are employed to ensure a fair trade-off between communication and sensing objectives. Numerical results demonstrate the superiority of the proposed beamforming design in terms of balancing communication and sensing performance.

The escalating interests on underwater exploration/reconnaissance applications have motivated high-rate data transmission from underwater to airborne relaying platforms, especially under high-sea scenarios. Thanks to its broad bandwidth and superior confidentiality, Optical wireless communication has become one promising candidate for water-air transmission. However, the optical signals inevitably suffer from deviations when crossing the highly-dynamic water-air interfaces in the absence of relaying ships/buoys. To address the issue, this article proposes one novel beam alignment strategy based on deep reinforcement learning (DRL) for water-air direct optical wireless communications. Specifically, the dynamic water-air interface is mathematically modeled using sea-wave spectrum analysis, followed by characterization of the propagation channel with ray-tracing techniques. Then the deep deterministic policy gradient (DDPG) scheme is introduced for DRL-based transceiving beam alignment. A logarithm-exponential (LE) nonlinear reward function with respect to the received signal strength is designed for high-resolution rewarding between different actions. Simulation results validate the superiority of the proposed DRL-based beam alignment scheme., Comment: 6 pages, 6 figures, published to ICCC 2024

Recent studies have shown that large language models' (LLMs) mathematical problem-solving capabilities can be enhanced by integrating external tools, such as code interpreters, and employing multi-turn Chain-of-Thought (CoT) reasoning. While current methods focus on synthetic data generation and Supervised Fine-Tuning (SFT), this paper studies the complementary direct preference learning approach to further improve model performance. However, existing direct preference learning algorithms are originally designed for the single-turn chat task, and do not fully address the complexities of multi-turn reasoning and external tool integration required for tool-integrated mathematical reasoning tasks. To fill in this gap, we introduce a multi-turn direct preference learning framework, tailored for this context, that leverages feedback from code interpreters and optimizes trajectory-level preferences. This framework includes multi-turn DPO and multi-turn KTO as specific implementations. The effectiveness of our framework is validated through training of various language models using an augmented prompt set from the GSM8K and MATH datasets. Our results demonstrate substantial improvements: a supervised fine-tuned Gemma-1.1-it-7B model's performance increased from 77.5% to 83.9% on GSM8K and from 46.1% to 51.2% on MATH. Similarly, a Gemma-2-it-9B model improved from 84.1% to 86.3% on GSM8K and from 51.0% to 54.5% on MATH., Comment: A multi-turn direct preference learning framework for tool-integrated reasoning tasks

Recently, the thermodynamics and kinetics of black hole phase transitions have garnered attention, particularly with the black hole radius being utilized as an order parameter within a generalized free energy landscape. In this framework, the local minima and maxima of the free energy correspond to stable and unstable states, respectively, while other states on the free energy landscape represent fluctuating black holes. Thermal fluctuations enable transitions between these stable states, and this stochastic kinetic behavior can be effectively described by the probabilistic Fokker-Planck equation.The transit time, defined as the time required for transitions or jumps between states, is a crucial physical quantity in phase transition kinetics, as it helps characterize the switching dynamics. By employing a harmonic transition state approximation, we examine the Hawking-Page phase transition and the Reissner-Nordstr\"om-Anti-de Sitter (RNAdS) black hole phase transition within the context of the generalized free energy landscape. We analytically quantify the transit time and its probability distribution for actual transition events, revealing the relationship between the mean transit time and the prefactor of the classical mean first passage time (MFPT). As the mean transit time decreases, the probability distribution narrows, indicating reduced fluctuations in transit time. These phenomena are connected to the topological structure of the generalized free energy landscape, the curvature of the free energy in both stable and unstable states, and the height of the energy barrier. We conclude that, for black hole phase transitions, the transit time serves as a characteristic timescale that reflects the actual jump time between states or to a transition state. Moreover, its relationship with the prefactor allows for a more precise quantification of the MFPT.

Temporal Knowledge Graphs (TKGs) incorporate temporal information to reflect the dynamic structural knowledge and evolutionary patterns of real-world facts. Nevertheless, TKGs are still limited in downstream applications due to the problem of incompleteness. Consequently, TKG completion (also known as link prediction) has been widely studied, with recent research focusing on incorporating independent embeddings of time or combining them with entities and relations to form temporal representations. However, most existing methods overlook the impact of history from a multi-granularity aspect. The inherent semantics of human-defined temporal granularities, such as ordinal dates, reveal general patterns to which facts typically adhere. To counter this limitation, this paper proposes \textbf{L}earning \textbf{G}ranularity \textbf{Re}presentation (termed $\mathsf{LGRe}$) for TKG completion. It comprises two main components: Granularity Representation Learning (GRL) and Adaptive Granularity Balancing (AGB). Specifically, GRL employs time-specific multi-layer convolutional neural networks to capture interactions between entities and relations at different granularities. After that, AGB generates adaptive weights for these embeddings according to temporal semantics, resulting in expressive representations of predictions. Moreover, to reflect similar semantics of adjacent timestamps, a temporal loss function is introduced. Extensive experimental results on four event benchmarks demonstrate the effectiveness of $\mathsf{LGRe}$ in learning time-related representations. To ensure reproducibility, our code is available at https://github.com/KcAcoZhang/LGRe., Comment: 15 pages. Accepted at ICONIP 2024

Plant disease recognition is a critical task that ensures crop health and mitigates the damage caused by diseases. A handy tool that enables farmers to receive a diagnosis based on query pictures or the text description of suspicious plants is in high demand for initiating treatment before potential diseases spread further. In this paper, we develop a multimodal plant disease image retrieval system to support disease search based on either image or text prompts. Specifically, we utilize the largest in-the-wild plant disease dataset PlantWild, which includes over 18,000 images across 89 categories, to provide a comprehensive view of potential diseases relating to the query. Furthermore, cross-modal retrieval is achieved in the developed system, facilitated by a novel CLIP-based vision-language model that encodes both disease descriptions and disease images into the same latent space. Built on top of the retriever, our retrieval system allows users to upload either plant disease images or disease descriptions to retrieve the corresponding images with similar characteristics from the disease dataset to suggest candidate diseases for end users' consideration.

With the increasing growth of industrialization, more and more industries are relying on machine automation for production. However, defect detection in large-scale production machinery is becoming increasingly important. Due to their large size and height, it is often challenging for professionals to conduct defect inspections on such large machinery. For example, the inspection of aging and misalignment of components on tall machinery like towers requires companies to assign dedicated personnel. Employees need to climb the towers and either visually inspect or take photos to detect safety hazards in these large machines. Direct visual inspection is limited by its low level of automation, lack of precision, and safety concerns associated with personnel climbing the towers. Therefore, in this paper, we propose a system based on neural network modeling (NeRF) of 3D twin models. By comparing two digital models, this system enables defect detection at the 3D interface of an object.

We introduce LAMPO, a novel paradigm that leverages Large Language Models (LLMs) for solving few-shot multi-class ordinal classification tasks. Unlike conventional methods, which concatenate all demonstration examples with the test instance and prompt LLMs to produce the pointwise prediction, our framework uses the LLM as a preference machine that makes a relative comparative decision between the test instance and each demonstration. A self-supervised method is then introduced to aggregate these binary comparisons into the final ordinal decision. LAMPO addresses several limitations inherent in previous methods, including context length constraints, ordering biases, and challenges associated with absolute point-wise estimation. Extensive experiments on seven public datasets demonstrate LAMPO's remarkably competitive performance across a diverse spectrum of applications (e.g., movie review analysis and hate speech detection). Notably, in certain applications, the improvement can be substantial, exceeding 20% in an absolute term. Moreover, we believe LAMPO represents an interesting addition to the non-parametric application layered on top of LLMs, as it supports black-box LLMs without necessitating the outputting of LLM's internal states (e.g., embeddings), as seen in previous approaches., Comment: COLM 2024

Existing plant disease classification models have achieved remarkable performance in recognizing in-laboratory diseased images. However, their performance often significantly degrades in classifying in-the-wild images. Furthermore, we observed that in-the-wild plant images may exhibit similar appearances across various diseases (i.e., small inter-class discrepancy) while the same diseases may look quite different (i.e., large intra-class variance). Motivated by this observation, we propose an in-the-wild multimodal plant disease recognition dataset that contains the largest number of disease classes but also text-based descriptions for each disease. Particularly, the newly provided text descriptions are introduced to provide rich information in textual modality and facilitate in-the-wild disease classification with small inter-class discrepancy and large intra-class variance issues. Therefore, our proposed dataset can be regarded as an ideal testbed for evaluating disease recognition methods in the real world. In addition, we further present a strong yet versatile baseline that models text descriptions and visual data through multiple prototypes for a given class. By fusing the contributions of multimodal prototypes in classification, our baseline can effectively address the small inter-class discrepancy and large intra-class variance issues. Remarkably, our baseline model can not only classify diseases but also recognize diseases in few-shot or training-free scenarios. Extensive benchmarking results demonstrate that our proposed in-the-wild multimodal dataset sets many new challenges to the plant disease recognition task and there is a large space to improve for future works.

Inspired by recent advances in Kolmogorov-Arnold Networks (KANs), we introduce a novel approach to latent factor conditional asset pricing models. While previous machine learning applications in asset pricing have predominantly used Multilayer Perceptrons with ReLU activation functions to model latent factor exposures, our method introduces a KAN-based autoencoder which surpasses MLP models in both accuracy and interpretability. Our model offers enhanced flexibility in approximating exposures as nonlinear functions of asset characteristics, while simultaneously providing users with an intuitive framework for interpreting latent factors. Empirical backtesting demonstrates our model's superior ability to explain cross-sectional risk exposures. Moreover, long-short portfolios constructed using our model's predictions achieve higher Sharpe ratios, highlighting its practical value in investment management., Comment: 7 pages

This study presents a comprehensive dataset capturing indoor environmental parameters, physiological responses, and subjective perceptions across three global cities. Utilizing wearable sensors, including smart eyeglasses, and a modified Cozie app, environmental and physiological data were collected, along with pre-screening, onboarding, and recurring surveys. Peripheral cues facilitated participant engagement with micro-EMA surveys, minimizing disruption over a 5-day collection period. The dataset offers insights into urban comfort dynamics, highlighting the interplay between environmental conditions, physiological responses, and subjective perceptions. Researchers can utilize this dataset to deepen their understanding of indoor environmental quality and inform the design of healthier built environments. Access to this dataset can advance indoor environmental research and contribute to the creation of more comfortable and sustainable indoor spaces., Comment: Submitted to Nature Scientific Data

Reward models (RMs) are crucial for aligning large language models (LLMs) with human preferences. They are trained using preference datasets where each example consists of one input prompt, two responses, and a preference label. As curating a high-quality human labeled preference dataset is both time-consuming and expensive, people often rely on existing powerful LLMs for preference label generation. This can potentially introduce noise and impede RM training. In this work, we present RMBoost, a novel synthetic preference data generation paradigm to boost reward model quality. Unlike traditional methods, which generate two responses before obtaining the preference label, RMBoost first generates one response and selects a preference label, followed by generating the second more (or less) preferred response conditioned on the pre-selected preference label and the first response. This approach offers two main advantages. First, RMBoost reduces labeling noise since preference pairs are constructed intentionally. Second, RMBoost facilitates the creation of more diverse responses by incorporating various quality aspects (e.g., helpfulness, relevance, completeness) into the prompts. We conduct extensive experiments across three diverse datasets and demonstrate that RMBoost outperforms other synthetic preference data generation techniques and significantly boosts the performance of four distinct reward models.

The popularity of automated news headline generation has surged with advancements in pre-trained language models. However, these models often suffer from the ``hallucination'' problem, where the generated headline is not fully supported by its source article. Efforts to address this issue have predominantly focused on English, using over-simplistic classification schemes that overlook nuanced hallucination types. In this study, we introduce the first multilingual, fine-grained news headline hallucination detection dataset that contains over 11 thousand pairs in 5 languages, each annotated with detailed hallucination types by experts. We conduct extensive experiments on this dataset under two settings. First, we implement several supervised fine-tuning approaches as preparatory solutions and demonstrate this dataset's challenges and utilities. Second, we test various large language models' in-context learning abilities and propose two novel techniques, language-dependent demonstration selection and coarse-to-fine prompting, to boost the few-shot hallucination detection performance in terms of the example-F1 metric. We release this dataset to foster further research in multilingual, fine-grained headline hallucination detection.

The identification of artefacts, particularly B-lines, in lung ultrasound (LUS), is crucial for assisting clinical diagnosis, prompting the development of innovative methodologies. While the Cauchy proximal splitting (CPS) algorithm has demonstrated effective performance in B-line detection, the process is slow and has limited generalization. This paper addresses these issues with a novel unsupervised deep unfolding network structure (DUCPS). The framework utilizes deep unfolding procedures to merge traditional model-based techniques with deep learning approaches. By unfolding the CPS algorithm into a deep network, DUCPS enables the parameters in the optimization algorithm to be learnable, thus enhancing generalization performance and facilitating rapid convergence. We conducted entirely unsupervised training using the Neighbor2Neighbor (N2N) and the Structural Similarity Index Measure (SSIM) losses. When combined with an improved line identification method proposed in this paper, state-of-the-art performance is achieved, with the recall and F2 score reaching 0.70 and 0.64, respectively. Notably, DUCPS significantly improves computational efficiency eliminating the need for extensive data labeling, representing a notable advancement over both traditional algorithms and existing deep learning approaches., Comment: 16 pages, 6 figures, IEEE TMI

Recently multi-channel end-to-end (ME2E) ASR systems have emerged. While streaming single-channel end-to-end ASR has been extensively studied, streaming ME2E ASR is limited in exploration. Additionally, recent studies call attention to the gap between in-distribution (ID) and out-of-distribution (OOD) tests and doing realistic evaluations. This paper focuses on two research problems: realizing streaming ME2E ASR and improving OOD generalization. We propose the CUSIDE-array method, which integrates the recent CUSIDE methodology (Chunking, Simulating Future Context and Decoding) into the neural beamformer approach of ME2E ASR. It enables streaming processing of both front-end and back-end with a total latency of 402ms. The CUSIDE-array ME2E models are shown to achieve superior streaming results in both ID and OOD tests. Realistic evaluations confirm the advantage of CUSIDE-array in its capability to consume single-channel data to improve OOD generalization via back-end pre-training and ME2E fine-tuning., Comment: Accepted into ISCSLP 2024

In the realm of self-supervised learning (SSL), masked image modeling (MIM) has gained popularity alongside contrastive learning methods. MIM involves reconstructing masked regions of input images using their unmasked portions. A notable subset of MIM methodologies employs discrete tokens as the reconstruction target, but the theoretical underpinnings of this choice remain underexplored. In this paper, we explore the role of these discrete tokens, aiming to unravel their benefits and limitations. Building upon the connection between MIM and contrastive learning, we provide a comprehensive theoretical understanding on how discrete tokenization affects the model's generalization capabilities. Furthermore, we propose a novel metric named TCAS, which is specifically designed to assess the effectiveness of discrete tokens within the MIM framework. Inspired by this metric, we contribute an innovative tokenizer design and propose a corresponding MIM method named ClusterMIM. It demonstrates superior performance on a variety of benchmark datasets and ViT backbones. Code is available at https://github.com/PKU-ML/ClusterMIM., Comment: ICLR 2024 Spotlight

The growing number of connected vehicles offers an opportunity to leverage internet of vehicles (IoV) data for traffic state estimation (TSE) which plays a crucial role in intelligent transportation systems (ITS). By utilizing only a portion of IoV data instead of the entire dataset, the significant overheads associated with collecting and processing large amounts of data can be avoided. In this paper, we introduce a novel framework that utilizes sparse IoV data to achieve cost-effective TSE. Particularly, we propose a novel spatial-temporal attention model called the convolutional retentive network (CRNet) to improve the TSE accuracy by mining spatial-temporal traffic state correlations. The model employs the convolutional neural network (CNN) for spatial correlation aggregation and the retentive network (RetNet) based on the attention mechanism to extract temporal correlations. Extensive simulations on a real-world IoV dataset validate the advantage of the proposed TSE approach in achieving accurate TSE using sparse IoV data, demonstrating its cost effectiveness and practicality for real-world applications., Comment: need further improvement

This paper introduces Standard Basis LoRA (SBoRA), a novel parameter-efficient fine-tuning approach for Large Language Models that builds upon the pioneering works of Low-Rank Adaptation (LoRA) and Orthogonal Adaptation. SBoRA reduces the number of trainable parameters by half or doubles the rank with the similar number of trainable parameters as LoRA, while improving learning performance. By utilizing orthogonal standard basis vectors to initialize one of the low-rank matrices (either $\mathbf{A}$ or $\mathbf{B}$), SBoRA facilitates regional weight updates and memory-efficient fine-tuning. This results in two variants, SBoRA-FA and SBoRA-FB, where only one of the matrices is updated, leading to a sparse update matrix $\mathrm{\Delta} \mathbf{W}$ with predominantly zero rows or columns. Consequently, most of the fine-tuned model's weights $(\mathbf{W}_0+\mathrm{\Delta} \mathbf{W})$ remain unchanged from the pre-trained weights, akin to the modular organization of the human brain, which efficiently adapts to new tasks. Our empirical results demonstrate the superiority of SBoRA-FA over LoRA in various fine-tuning tasks, including commonsense reasoning and arithmetic reasoning. Furthermore, we evaluate the effectiveness of QSBoRA on quantized LLaMA models of varying scales, highlighting its potential for efficient adaptation to new tasks. Code is available at https://github.com/cityuhkai/SBoRA, Comment: 16 pages, 4 figures

Existing human datasets for avatar creation are typically limited to laboratory environments, wherein high-quality annotations (e.g., SMPL estimation from 3D scans or multi-view images) can be ideally provided. However, their annotating requirements are impractical for real-world images or videos, posing challenges toward real-world applications on current avatar creation methods. To this end, we propose the WildAvatar dataset, a web-scale in-the-wild human avatar creation dataset extracted from YouTube, with $10,000+$ different human subjects and scenes. WildAvatar is at least $10\times$ richer than previous datasets for 3D human avatar creation. We evaluate several state-of-the-art avatar creation methods on our dataset, highlighting the unexplored challenges in real-world applications on avatar creation. We also demonstrate the potential for generalizability of avatar creation methods, when provided with data at scale. We publicly release our data source links and annotations, to push forward 3D human avatar creation and other related fields for real-world applications., Comment: Project page: https://wildavatar.github.io/

The development of autonomous agents increasingly relies on Multimodal Language Models (MLMs) to perform tasks described in natural language with GUI environments, such as websites, desktop computers, or mobile phones. Existing benchmarks for MLM agents in interactive environments are limited by their focus on a single environment, lack of detailed and generalized evaluation methods, and the complexities of constructing tasks and evaluators. To overcome these limitations, we introduce Crab, the first agent benchmark framework designed to support cross-environment tasks, incorporating a graph-based fine-grained evaluation method and an efficient mechanism for task and evaluator construction. Our framework supports multiple devices and can be easily extended to any environment with a Python interface. Leveraging Crab, we developed a cross-platform Crab Benchmark-v0 comprising 120 tasks in computer desktop and mobile phone environments. We evaluated four advanced MLMs using different single and multi-agent system configurations on this benchmark. The experimental results demonstrate that the single agent with GPT-4o achieves the best completion ratio of 38.01%. All framework code, agent code, and task datasets are publicly available at https://github.com/camel-ai/crab.

In recent years, the rise of generative self-supervised learning (SSL) paradigms has exhibited impressive performance across visual, language, and multi-modal domains. While the varied designs of generative SSL objectives lead to distinct properties in downstream tasks, a theoretical understanding of these differences remains largely unexplored. In this paper, we establish the first theoretical comparisons between two leading generative SSL paradigms: autoregressive SSL and masked SSL. Through establishing theoretical frameworks, we elucidate the strengths and limitations of autoregressive and masked SSL within the primary evaluation tasks of classification and content generation. Our findings demonstrate that in classification tasks, the flexibility of targeted tokens in masked SSL fosters more inter-sample connections compared to the fixed position of target tokens in autoregressive SSL, which yields superior clustering performance. In content generation tasks, the misalignment between the flexible lengths of test samples and the fixed length of unmasked texts in masked SSL (vs. flexible lengths of conditional texts in autoregressive SSL) hinders its generation performance. To leverage each other's strengths and mitigate weaknesses, we propose diversity-enhanced autoregressive and variable-length masked objectives, which substantially improve the classification performance of autoregressive SSL and the generation performance of masked SSL. Code is available at https://github.com/PKU-ML/LookAheadLookAround.