While Large Language Models (LLMs) have demonstrated exceptional multitasking abilities, fine-tuning these models on downstream, domain-specific datasets is often necessary to yield superior performance on test sets compared to their counterparts without fine-tuning. However, the comprehensive effects of fine-tuning on the LLMs' generalization ability are not fully understood. This paper delves into the differences between original, unmodified LLMs and their fine-tuned variants. Our primary investigation centers on whether fine-tuning affects the generalization ability intrinsic to LLMs. To elaborate on this, we conduct extensive experiments across five distinct language tasks on various datasets. Our main findings reveal that models fine-tuned on generation and classification tasks exhibit dissimilar behaviors in generalizing to different domains and tasks. Intriguingly, we observe that integrating the in-context learning strategy during fine-tuning on generation tasks can enhance the model's generalization ability. Through this systematic investigation, we aim to contribute valuable insights into the evolving landscape of fine-tuning practices for LLMs.
相關內容
According to the World Health Organization, the involvement of Vulnerable Road Users (VRUs) in traffic accidents remains a significant concern, with VRUs accounting for over half of traffic fatalities. The increase of automation and connectivity levels of vehicles has still an uncertain impact on VRU safety. By deploying the Collective Perception Service (CPS), vehicles can include information about VRUs in Vehicle-to-Everything (V2X) messages, thus raising the general perception of the environment. Although an increased awareness is considered positive, one could argue that the awareness ratio, the metric used to measure perception, is only implicitly connected to the VRUs' safety. This paper introduces a tailored metric, the Risk Factor (RF), to measure the risk level for the interactions between Connected Automated Vehicles (CAVs) and VRUs. By evaluating the RF, we assess the impact of V2X communication on VRU risk mitigation. Our results show that high V2X penetration rates can reduce mean risk, quantified by our proposed metric, by up to 44%. Although the median risk value shows a significant decrease, suggesting a reduction in overall risk, the distribution of risk values reveals that CPS's mitigation effectiveness is overestimated, which is indicated by the divergence between RF and awareness ratio. Additionally, by analyzing a real-world traffic dataset, we pinpoint high-risk locations within a scenario, identifying areas near intersections and behind parked cars as especially dangerous. Our methodology can be ported and applied to other scenarios in order to identify high-risk areas. We value the proposed RF as an insightful metric for quantifying VRU safety in a highly automated and connected environment.
One key aspect differentiating data-driven single- and multi-channel speech enhancement and dereverberation methods is that both the problem formulation and complexity of the solutions are considerably more challenging in the latter case. Additionally, with limited computational resources, it is cumbersome to train models that require the management of larger datasets or those with more complex designs. In this scenario, an unverified hypothesis that single-channel methods can be adapted to multi-channel scenarios simply by processing each channel independently holds significant implications, boosting compatibility between sound scene capture and system input-output formats, while also allowing modern research to focus on other challenging aspects, such as full-bandwidth audio enhancement, competitive noise suppression, and unsupervised learning. This study verifies this hypothesis by comparing the enhancement promoted by a basic single-channel speech enhancement and dereverberation model with two other multi-channel models tailored to separate clean speech from noisy 3D mixes. A direction of arrival estimation model was used to objectively evaluate its capacity to preserve spatial information by comparing the output signals with ground-truth coordinate values. Consequently, a trade-off arises between preserving spatial information with a more straightforward single-channel solution at the cost of obtaining lower gains in intelligibility scores.
The burgeoning interest in developing Large Language Models (LLMs) with up to trillion parameters has been met with concerns regarding resource efficiency and practical expense, particularly given the immense cost of experimentation. This scenario underscores the importance of exploring the potential of Small Language Models (SLMs) as a resource-efficient alternative. In this context, we introduce MiniCPM, specifically the 1.2B and 2.4B non-embedding parameter variants, not only excel in their respective categories but also demonstrate capabilities on par with 7B-13B LLMs. While focusing on SLMs, our approach exhibits scalability in both model and data dimensions for future LLM research. Regarding model scaling, we employ extensive model wind tunnel experiments for stable and optimal scaling. For data scaling, we introduce a Warmup-Stable-Decay (WSD) learning rate scheduler (LRS), conducive to continuous training and domain adaptation. We present an in-depth analysis of the intriguing training dynamics that occurred in the WSD LRS. With WSD LRS, we are now able to efficiently study data-model scaling law without extensive retraining experiments on both axes of model and data, from which we derive the much higher compute optimal data-model ratio than Chinchilla Optimal. Additionally, we introduce MiniCPM family, including MiniCPM-DPO, MiniCPM-MoE and MiniCPM-128K, whose excellent performance further cementing MiniCPM's foundation in diverse SLM applications. MiniCPM models are available publicly at //github.com/OpenBMB/MiniCPM .
Large Language Models (LLMs) have significantly impacted the writing process, enabling collaborative content creation and enhancing productivity. However, generating high-quality, user-aligned text remains challenging. In this paper, we propose Writing Path, a framework that uses explicit outlines to guide LLMs in generating goal-oriented, high-quality pieces of writing. Our approach draws inspiration from structured writing planning and reasoning paths, focusing on capturing and reflecting user intentions throughout the writing process. We construct a diverse dataset from unstructured blog posts to benchmark writing performance and introduce a comprehensive evaluation framework assessing the quality of outlines and generated texts. Our evaluations with GPT-3.5-turbo, GPT-4, and HyperCLOVA X demonstrate that the Writing Path approach significantly enhances text quality according to both LLMs and human evaluations. This study highlights the potential of integrating writing-specific techniques into LLMs to enhance their ability to meet the diverse writing needs of users.
Large Vision-Language Models (LVLMs) suffer from hallucination issues, wherein the models generate plausible-sounding but factually incorrect outputs, undermining their reliability. A comprehensive quantitative evaluation is necessary to identify and understand the extent of hallucinations in these models. However, existing benchmarks are often limited in scope, focusing mainly on object hallucinations. Furthermore, current evaluation methods struggle to effectively address the subtle semantic distinctions between model outputs and reference data, as well as the balance between hallucination and informativeness. To address these issues, we introduce a multi-dimensional benchmark covering objects, attributes, and relations, with challenging images selected based on associative biases. Moreover, we propose an large language model (LLM)-based two-stage evaluation framework that generalizes the popular CHAIR metric and incorporates both faithfulness and coverage into the evaluation. Experiments on 10 established LVLMs demonstrate that our evaluation metric is more comprehensive and better correlated with humans than existing work when evaluating on our challenging human annotated benchmark dataset. Our work also highlights the critical balance between faithfulness and coverage of model outputs, and encourages future works to address hallucinations in LVLMs while keeping their outputs informative.
With the rapid advancements in Multimodal Large Language Models (MLLMs), securing these models against malicious inputs while aligning them with human values has emerged as a critical challenge. In this paper, we investigate an important and unexplored question of whether techniques that successfully jailbreak Large Language Models (LLMs) can be equally effective in jailbreaking MLLMs. To explore this issue, we introduce JailBreakV-28K, a pioneering benchmark designed to assess the transferability of LLM jailbreak techniques to MLLMs, thereby evaluating the robustness of MLLMs against diverse jailbreak attacks. Utilizing a dataset of 2, 000 malicious queries that is also proposed in this paper, we generate 20, 000 text-based jailbreak prompts using advanced jailbreak attacks on LLMs, alongside 8, 000 image-based jailbreak inputs from recent MLLMs jailbreak attacks, our comprehensive dataset includes 28, 000 test cases across a spectrum of adversarial scenarios. Our evaluation of 10 open-source MLLMs reveals a notably high Attack Success Rate (ASR) for attacks transferred from LLMs, highlighting a critical vulnerability in MLLMs that stems from their text-processing capabilities. Our findings underscore the urgent need for future research to address alignment vulnerabilities in MLLMs from both textual and visual inputs.
Hallucination in Natural Language Generation (NLG) is like the elephant in the room, obvious but often overlooked until recent achievements significantly improved the fluency and grammatical accuracy of generated text. For Large Language Models (LLMs), hallucinations can happen in various downstream tasks and casual conversations, which need accurate assessment to enhance reliability and safety. However, current studies on hallucination evaluation vary greatly, and people still find it difficult to sort out and select the most appropriate evaluation methods. Moreover, as NLP research gradually shifts to the domain of LLMs, it brings new challenges to this direction. This paper provides a comprehensive survey on the evolvement of hallucination evaluation methods, aiming to address three key aspects: 1) Diverse definitions and granularity of facts; 2) The categories of automatic evaluators and their applicability; 3) Unresolved issues and future directions.
Latent Diffusion Models (LDMs) have emerged as powerful generative models, known for delivering remarkable results under constrained computational resources. However, deploying LDMs on resource-limited devices remains a complex issue, presenting challenges such as memory consumption and inference speed. To address this issue, we introduce LD-Pruner, a novel performance-preserving structured pruning method for compressing LDMs. Traditional pruning methods for deep neural networks are not tailored to the unique characteristics of LDMs, such as the high computational cost of training and the absence of a fast, straightforward and task-agnostic method for evaluating model performance. Our method tackles these challenges by leveraging the latent space during the pruning process, enabling us to effectively quantify the impact of pruning on model performance, independently of the task at hand. This targeted pruning of components with minimal impact on the output allows for faster convergence during training, as the model has less information to re-learn, thereby addressing the high computational cost of training. Consequently, our approach achieves a compressed model that offers improved inference speed and reduced parameter count, while maintaining minimal performance degradation. We demonstrate the effectiveness of our approach on three different tasks: text-to-image (T2I) generation, Unconditional Image Generation (UIG) and Unconditional Audio Generation (UAG). Notably, we reduce the inference time of Stable Diffusion (SD) by 34.9% while simultaneously improving its FID by 5.2% on MS-COCO T2I benchmark. This work paves the way for more efficient pruning methods for LDMs, enhancing their applicability.
Multi-modal AI systems will likely become a ubiquitous presence in our everyday lives. A promising approach to making these systems more interactive is to embody them as agents within physical and virtual environments. At present, systems leverage existing foundation models as the basic building blocks for the creation of embodied agents. Embedding agents within such environments facilitates the ability of models to process and interpret visual and contextual data, which is critical for the creation of more sophisticated and context-aware AI systems. For example, a system that can perceive user actions, human behavior, environmental objects, audio expressions, and the collective sentiment of a scene can be used to inform and direct agent responses within the given environment. To accelerate research on agent-based multimodal intelligence, we define "Agent AI" as a class of interactive systems that can perceive visual stimuli, language inputs, and other environmentally-grounded data, and can produce meaningful embodied action with infinite agent. In particular, we explore systems that aim to improve agents based on next-embodied action prediction by incorporating external knowledge, multi-sensory inputs, and human feedback. We argue that by developing agentic AI systems in grounded environments, one can also mitigate the hallucinations of large foundation models and their tendency to generate environmentally incorrect outputs. The emerging field of Agent AI subsumes the broader embodied and agentic aspects of multimodal interactions. Beyond agents acting and interacting in the physical world, we envision a future where people can easily create any virtual reality or simulated scene and interact with agents embodied within the virtual environment.
Deep Convolutional Neural Networks (CNNs) are a special type of Neural Networks, which have shown state-of-the-art results on various competitive benchmarks. The powerful learning ability of deep CNN is largely achieved with the use of multiple non-linear feature extraction stages that can automatically learn hierarchical representation from the data. Availability of a large amount of data and improvements in the hardware processing units have accelerated the research in CNNs and recently very interesting deep CNN architectures are reported. The recent race in deep CNN architectures for achieving high performance on the challenging benchmarks has shown that the innovative architectural ideas, as well as parameter optimization, can improve the CNN performance on various vision-related tasks. In this regard, different ideas in the CNN design have been explored such as use of different activation and loss functions, parameter optimization, regularization, and restructuring of processing units. However, the major improvement in representational capacity is achieved by the restructuring of the processing units. Especially, the idea of using a block as a structural unit instead of a layer is gaining substantial appreciation. This survey thus focuses on the intrinsic taxonomy present in the recently reported CNN architectures and consequently, classifies the recent innovations in CNN architectures into seven different categories. These seven categories are based on spatial exploitation, depth, multi-path, width, feature map exploitation, channel boosting and attention. Additionally, it covers the elementary understanding of the CNN components and sheds light on the current challenges and applications of CNNs.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191