Video Large Language Models (Video-LLMs) are flourishing and has advanced many video-language tasks. As a golden testbed, Video Question Answering (VideoQA) plays pivotal role in Video-LLM developing. This work conducts a timely and comprehensive study of Video-LLMs' behavior in VideoQA, aiming to elucidate their success and failure modes, and provide insights towards more human-like video understanding and question answering. Our analyses demonstrate that Video-LLMs excel in VideoQA; they can correlate contextual cues and generate plausible responses to questions about varied video contents. However, models falter in handling video temporality, both in reasoning about temporal content ordering and grounding QA-relevant temporal moments. Moreover, the models behave unintuitively - they are unresponsive to adversarial video perturbations while being sensitive to simple variations of candidate answers and questions. Also, they do not necessarily generalize better. The findings demonstrate Video-LLMs' QA capability in standard condition yet highlight their severe deficiency in robustness and interpretability, suggesting the urgent need on rationales in Video-LLM developing.
相關內容
Large Language Models have demonstrated remarkable abilities across various tasks, with Chain-of-Thought (CoT) prompting emerging as a key technique to enhance reasoning capabilities. However, existing research primarily focuses on improving performance, lacking a comprehensive framework to explain and understand the fundamental factors behind CoT's success. To bridge this gap, we introduce a novel perspective grounded in the Hopfieldian view of cognition in cognitive neuroscience. We establish a connection between CoT reasoning and key cognitive elements such as stimuli, actions, neural populations, and representation spaces. From our view, we can understand the reasoning process as the movement between these representation spaces. Building on this insight, we develop a method for localizing reasoning errors in the response of CoTs. Moreover, we propose the Representation-of-Thought (RoT) framework, which leverages the robustness of low-dimensional representation spaces to enhance the robustness of the reasoning process in CoTs. Experimental results demonstrate that RoT improves the robustness and interpretability of CoT reasoning while offering fine-grained control over the reasoning process.
Large Language Models (LLMs) have shown impressive abilities in code generation, but they may generate erroneous programs. Reading a program takes ten times longer than writing it. Showing these erroneous programs to developers will waste developers' energies and introduce security risks to software. To address the above limitations, we propose HonestCoder, a novel LLM-based code generation approach. HonestCoder selectively shows the generated programs to developers based on LLMs' confidence. The confidence provides valuable insights into the correctness of generated programs. To achieve this goal, we propose a novel approach to estimate LLMs' confidence in code generation. It estimates confidence by measuring the multi-modal similarity between LLMs-generated programs. We collect and release a multilingual benchmark named TruthCodeBench, which consists of 2,265 samples and covers two popular programming languages (i.e., Python and Java). We apply HonestCoder to four popular LLMs (e.g., DeepSeek-Coder and Code Llama) and evaluate it on TruthCodeBench. Based on the experiments, we obtain the following insights. (1) HonestCoder can effectively estimate LLMs' confidence and accurately determine the correctness of generated programs. For example, HonestCoder outperforms the state-of-the-art baseline by 27.79% in AUROC and 63.74% in AUCPR. (2) HonestCoder can decrease the number of erroneous programs shown to developers. Compared to eight baselines, it can show more correct programs and fewer erroneous programs to developers. (3) Compared to showing code indiscriminately, HonestCoder only adds slight time overhead (approximately 0.4 seconds per requirement). (4) We discuss future directions to facilitate the application of LLMs in software development. We hope this work can motivate broad discussions about measuring the reliability of LLMs' outputs in performing code-related tasks.
Video Large Language Models (Video-LLMs) have made remarkable progress in video understanding tasks. However, they are constrained by the maximum length of input tokens, making it impractical to input entire videos. Existing frame selection approaches, such as uniform frame sampling and text-frame retrieval, fail to account for the information density variations in the videos or the complex instructions in the tasks, leading to sub-optimal performance. In this paper, we propose Frame-Voyager that learns to query informative frame combinations, based on the given textual queries in the task. To train Frame-Voyager, we introduce a new data collection and labeling pipeline, by ranking frame combinations using a pre-trained Video-LLM. Given a video of M frames, we traverse its T-frame combinations, feed them into a Video-LLM, and rank them based on Video-LLM's prediction losses. Using this ranking as supervision, we train Frame-Voyager to query the frame combinations with lower losses. In experiments, we evaluate Frame-Voyager on four Video Question Answering benchmarks by plugging it into two different Video-LLMs. The experimental results demonstrate that Frame-Voyager achieves impressive results in all settings, highlighting its potential as a plug-and-play solution for Video-LLMs.
Although Large Language Models (LLMs) have demonstrated significant capabilities in executing complex tasks in a zero-shot manner, they are susceptible to jailbreak attacks and can be manipulated to produce harmful outputs. Recently, a growing body of research has categorized jailbreak attacks into token-level and prompt-level attacks. However, previous work primarily overlooks the diverse key factors of jailbreak attacks, with most studies concentrating on LLM vulnerabilities and lacking exploration of defense-enhanced LLMs. To address these issues, we evaluate the impact of various attack settings on LLM performance and provide a baseline benchmark for jailbreak attacks, encouraging the adoption of a standardized evaluation framework. Specifically, we evaluate the eight key factors of implementing jailbreak attacks on LLMs from both target-level and attack-level perspectives. We further conduct seven representative jailbreak attacks on six defense methods across two widely used datasets, encompassing approximately 354 experiments with about 55,000 GPU hours on A800-80G. Our experimental results highlight the need for standardized benchmarking to evaluate these attacks on defense-enhanced LLMs. Our code is available at //github.com/usail-hkust/Bag_of_Tricks_for_LLM_Jailbreaking.
We propose a novel method for reconstructing explicit parameterized surfaces from Signed Distance Fields (SDFs), a widely used implicit neural representation (INR) for 3D surfaces. While traditional reconstruction methods like Marching Cubes extract discrete meshes that lose the continuous and differentiable properties of INRs, our approach iteratively contracts a parameterized initial sphere to conform to the target SDF shape, preserving differentiability and surface parameterization throughout. This enables downstream applications such as texture mapping, geometry processing, animation, and finite element analysis. Evaluated on the typical geometric shapes and parts of the ABC dataset, our method achieves competitive reconstruction quality, maintaining smoothness and differentiability crucial for advanced computer graphics and geometric deep learning applications.
Over the past decade, Graph Neural Networks (GNNs) have transformed graph representation learning. In the widely adopted message-passing GNN framework, nodes refine their representations by aggregating information from neighboring nodes iteratively. While GNNs excel in various domains, recent theoretical studies have raised concerns about their capabilities. GNNs aim to address various graph-related tasks by utilizing such node representations, however, this one-size-fits-all approach proves suboptimal for diverse tasks. Motivated by these observations, we conduct empirical tests to compare the performance of current GNN models with more conventional and direct methods in link prediction tasks. Introducing our model, PROXI, which leverages proximity information of node pairs in both graph and attribute spaces, we find that standard machine learning (ML) models perform competitively, even outperforming cutting-edge GNN models when applied to these proximity metrics derived from node neighborhoods and attributes. This holds true across both homophilic and heterophilic networks, as well as small and large benchmark datasets, including those from the Open Graph Benchmark (OGB). Moreover, we show that augmenting traditional GNNs with PROXI significantly boosts their link prediction performance. Our empirical findings corroborate the previously mentioned theoretical observations and imply that there exists ample room for enhancement in current GNN models to reach their potential.
Large Language Models (LLMs) show remarkable performance on a wide variety of tasks. Most LLMs split text into multi-character tokens and process them as atomic units without direct access to individual characters. This raises the question: To what extent can LLMs learn orthographic information? To answer this, we propose a new benchmark, CUTE, which features a collection of tasks designed to test the orthographic knowledge of LLMs. We evaluate popular LLMs on CUTE, finding that most of them seem to know the spelling of their tokens, yet fail to use this information effectively to manipulate text, calling into question how much of this knowledge is generalizable.
Dealing with Partially Observable Markov Decision Processes is notably a challenging task. We face an average-reward infinite-horizon POMDP setting with an unknown transition model, where we assume the knowledge of the observation model. Under this assumption, we propose the Observation-Aware Spectral (OAS) estimation technique, which enables the POMDP parameters to be learned from samples collected using a belief-based policy. Then, we propose the OAS-UCRL algorithm that implicitly balances the exploration-exploitation trade-off following the $\textit{optimism in the face of uncertainty}$ principle. The algorithm runs through episodes of increasing length. For each episode, the optimal belief-based policy of the estimated POMDP interacts with the environment and collects samples that will be used in the next episode by the OAS estimation procedure to compute a new estimate of the POMDP parameters. Given the estimated model, an optimization oracle computes the new optimal policy. We show the consistency of the OAS procedure, and we prove a regret guarantee of order $\mathcal{O}(\sqrt{T \log(T)})$ for the proposed OAS-UCRL algorithm. We compare against the oracle playing the optimal stochastic belief-based policy and show the efficient scaling of our approach with respect to the dimensionality of the state, action, and observation space. We finally conduct numerical simulations to validate and compare the proposed technique with other baseline approaches.
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.
Knowledge plays a critical role in artificial intelligence. Recently, the extensive success of pre-trained language models (PLMs) has raised significant attention about how knowledge can be acquired, maintained, updated and used by language models. Despite the enormous amount of related studies, there still lacks a unified view of how knowledge circulates within language models throughout the learning, tuning, and application processes, which may prevent us from further understanding the connections between current progress or realizing existing limitations. In this survey, we revisit PLMs as knowledge-based systems by dividing the life circle of knowledge in PLMs into five critical periods, and investigating how knowledge circulates when it is built, maintained and used. To this end, we systematically review existing studies of each period of the knowledge life cycle, summarize the main challenges and current limitations, and discuss future directions.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191