Abstract:The scaling of large language models (LLMs) is a critical research area for the efficiency and effectiveness of model training and deployment. Our work investigates the transferability and discrepancies of scaling laws between Dense Models and Mixture of Experts (MoE) models. Through a combination of theoretical analysis and extensive experiments, including consistent loss scaling, optimal batch size and learning rate scaling, and resource allocation strategies scaling, our findings reveal that the power-law scaling framework also applies to MoE Models, indicating that the fundamental principles governing the scaling behavior of these models are preserved, even though the architecture differs. Additionally, MoE Models demonstrate superior generalization, resulting in lower testing losses with the same training compute budget compared to Dense Models. These findings indicate the scaling consistency and transfer generalization capabilities of MoE Models, providing new insights for optimizing MoE Model training and deployment strategies.
Abstract:Artificial intelligence (AI) plays a crucial role in autonomous driving (AD) research, propelling its development towards intelligence and efficiency. Currently, the development of AD technology follows two main technical paths: modularization and end-to-end. Modularization decompose the driving task into modules such as perception, prediction, planning, and control, and train them separately. Due to the inconsistency of training objectives between modules, the integrated effect suffers from bias. End-to-end attempts to address this issue by utilizing a single model that directly maps from sensor data to control signals. This path has limited learning capabilities in a comprehensive set of features and struggles to handle unpredictable long-tail events and complex urban traffic scenarios. In the face of challenges encountered in both paths, many researchers believe that large language models (LLMs) with powerful reasoning capabilities and extensive knowledge understanding may be the solution, expecting LLMs to provide AD systems with deeper levels of understanding and decision-making capabilities. In light of the challenges faced by both paths, many researchers believe that LLMs, with their powerful reasoning abilities and extensive knowledge, could offer a solution. To understand if LLMs could enhance AD, this paper conducts a thorough analysis of the potential applications of LLMs in AD systems, including exploring their optimization strategies in both modular and end-to-end approaches, with a particular focus on how LLMs can tackle the problems and challenges present in current solutions. Furthermore, we discuss an important question: Can LLM-based artificial general intelligence (AGI) be a key to achieve high-level AD? We further analyze the potential limitations and challenges that LLMs may encounter in promoting the development of AD technology.
Abstract:Vertical federated learning is a natural and elegant approach to integrate multi-view data vertically partitioned across devices (clients) while preserving their privacies. Apart from the model training, existing methods requires the collaboration of all clients in the model inference. However, the model inference is probably maintained for service in a long time, while the collaboration, especially when the clients belong to different organizations, is unpredictable in real-world scenarios, such as concellation of contract, network unavailablity, etc., resulting in the failure of them. To address this issue, we, at the first attempt, propose a flexible Active-Passive Federated learning (APFed) framework. Specifically, the active client is the initiator of a learning task and responsible to build the complete model, while the passive clients only serve as assistants. Once the model built, the active client can make inference independently. In addition, we instance the APFed framework into two classification methods with employing the reconstruction loss and the contrastive loss on passive clients, respectively. Meanwhile, the two methods are tested in a set of experiments and achieves desired results, validating their effectiveness.
Abstract:In e-commerce, the order in which search results are displayed when a customer tries to find relevant listings can significantly impact their shopping experience and search efficiency. Tailored re-ranking system based on relevance and engagement signals in E-commerce has often shown improvement on sales and gross merchandise value (GMV). Designing algorithms for this purpose is even more challenging when the shops are not restricted to domestic buyers, but can sale globally to international buyers. Our solution needs to incorporate shopping preference and cultural traditions in different buyer markets. We propose the SEQ+MD framework, which integrates sequential learning for multi-task learning (MTL) and feature-generated region-mask for multi-distribution input. This approach leverages the sequential order within tasks and accounts for regional heterogeneity, enhancing performance on multi-source data. Evaluations on in-house data showed a strong increase on the high-value engagement including add-to-cart and purchase while keeping click performance neutral compared to state-of-the-art baseline models. Additionally, our multi-regional learning module is "plug-and-play" and can be easily adapted to enhance other MTL applications.
Abstract:As large language models (LLMs) constantly evolve, ensuring their safety remains a critical research problem. Previous red-teaming approaches for LLM safety have primarily focused on single prompt attacks or goal hijacking. To the best of our knowledge, we are the first to study LLM safety in multi-turn dialogue coreference. We created a dataset of 1,400 questions across 14 categories, each featuring multi-turn coreference safety attacks. We then conducted detailed evaluations on five widely used open-source LLMs. The results indicated that under multi-turn coreference safety attacks, the highest attack success rate was 56% with the LLaMA2-Chat-7b model, while the lowest was 13.9% with the Mistral-7B-Instruct model. These findings highlight the safety vulnerabilities in LLMs during dialogue coreference interactions.
Abstract:Multimodal Entity Linking (MEL) aims to link ambiguous mentions in multimodal contexts to entities in a multimodal knowledge graph. A pivotal challenge is to fully leverage multi-element correlations between mentions and entities to bridge modality gap and enable fine-grained semantic matching. Existing methods attempt several local correlative mechanisms, relying heavily on the automatically learned attention weights, which may over-concentrate on partial correlations. To mitigate this issue, we formulate the correlation assignment problem as an optimal transport (OT) problem, and propose a novel MEL framework, namely OT-MEL, with OT-guided correlation assignment. Thereby, we exploit the correlation between multimodal features to enhance multimodal fusion, and the correlation between mentions and entities to enhance fine-grained matching. To accelerate model prediction, we further leverage knowledge distillation to transfer OT assignment knowledge to attention mechanism. Experimental results show that our model significantly outperforms previous state-of-the-art baselines and confirm the effectiveness of the OT-guided correlation assignment.
Abstract:Text-centric visual question answering (VQA) has made great strides with the development of Multimodal Large Language Models (MLLMs), yet open-source models still fall short of leading models like GPT4V and Gemini, partly due to a lack of extensive, high-quality instruction tuning data. To this end, we introduce a new approach for creating a massive, high-quality instruction-tuning dataset, Square-10M, which is generated using closed-source MLLMs. The data construction process, termed Square, consists of four steps: Self-Questioning, Answering, Reasoning, and Evaluation. Our experiments with Square-10M led to three key findings: 1) Our model, TextSquare, considerably surpasses open-source previous state-of-the-art Text-centric MLLMs and sets a new standard on OCRBench(62.2%). It even outperforms top-tier models like GPT4V and Gemini in 6 of 10 text-centric benchmarks. 2) Additionally, we demonstrate the critical role of VQA reasoning data in offering comprehensive contextual insights for specific questions. This not only improves accuracy but also significantly mitigates hallucinations. Specifically, TextSquare scores an average of 75.1% across four general VQA and hallucination evaluation datasets, outperforming previous state-of-the-art models. 3) Notably, the phenomenon observed in scaling text-centric VQA datasets reveals a vivid pattern: the exponential increase of instruction tuning data volume is directly proportional to the improvement in model performance, thereby validating the necessity of the dataset scale and the high quality of Square-10M.
Abstract:Digital Image Correlation (DIC) is an optical technique that measures displacement and strain by tracking pattern movement in a sequence of captured images during testing. DIC has gained recognition in asphalt pavement engineering since the early 2000s. However, users often perceive the DIC technique as an out-of-box tool and lack a thorough understanding of its operational and measurement principles. This article presents a state-of-art review of DIC as a crucial tool for laboratory testing of asphalt concrete (AC), primarily focusing on the widely utilized 2D-DIC and 3D-DIC techniques. To address frequently asked questions from users, the review thoroughly examines the optimal methods for preparing speckle patterns, configuring single-camera or dual-camera imaging systems, conducting DIC analyses, and exploring various applications. Furthermore, emerging DIC methodologies such as Digital Volume Correlation and deep-learning-based DIC are introduced, highlighting their potential for future applications in pavement engineering. The article also provides a comprehensive and reliable flowchart for implementing DIC in AC characterization. Finally, critical directions for future research are presented.
Abstract:The conventional surface reflection method has been widely used to measure the asphalt pavement layer dielectric constant using ground-penetrating radar (GPR). This method may be inaccurate for in-service pavement thickness estimation with dielectric constant variation through the depth, which could be addressed using the extended common mid-point method (XCMP) with air-coupled GPR antennas. However, the factors affecting the XCMP method on thickness prediction accuracy haven't been studied. Manual acquisition of key factors is required, which hinders its real-time applications. This study investigates the affecting factors and develops a modified XCMP method to allow automatic thickness prediction of in-service asphalt pavement with non-uniform dielectric properties through depth. A sensitivity analysis was performed, necessitating the accurate estimation of time of flights (TOFs) from antenna pairs. A modified XCMP method based on edge detection was proposed to allow real-time TOFs estimation, then dielectric constant and thickness predictions. Field tests using a multi-channel GPR system were performed for validation. Both the surface reflection and XCMP setups were conducted. Results show that the modified XCMP method is recommended with a mean prediction error of 1.86%, which is more accurate than the surface reflection method (5.73%).
Abstract:The widespread of generative artificial intelligence has heightened concerns about the potential harms posed by AI-generated texts, primarily stemming from factoid, unfair, and toxic content. Previous researchers have invested much effort in assessing the harmlessness of generative language models. However, existing benchmarks are struggling in the era of large language models (LLMs), due to the stronger language generation and instruction following capabilities, as well as wider applications. In this paper, we propose FFT, a new benchmark with 2116 elaborated-designed instances, for LLM harmlessness evaluation with factuality, fairness, and toxicity. To investigate the potential harms of LLMs, we evaluate 9 representative LLMs covering various parameter scales, training stages, and creators. Experiments show that the harmlessness of LLMs is still under-satisfactory, and extensive analysis derives some insightful findings that could inspire future research for harmless LLM research.