Human values and their measurement are long-standing interdisciplinary inquiry. Recent advances in AI have sparked renewed interest in this area, with large language models (LLMs) emerging as both tools and subjects of value measurement. This work introduces Generative Psychometrics for Values (GPV), an LLM-based, data-driven value measurement paradigm, theoretically grounded in text-revealed selective perceptions. The core idea is to dynamically parse unstructured texts into perceptions akin to static stimuli in traditional psychometrics, measure the value orientations they reveal, and aggregate the results. Applying GPV to human-authored blogs, we demonstrate its stability, validity, and superiority over prior psychological tools. Then, extending GPV to LLM value measurement, we advance the current art with 1) a psychometric methodology that measures LLM values based on their scalable and free-form outputs, enabling context-specific measurement; 2) a comparative analysis of measurement paradigms, indicating response biases of prior methods; and 3) an attempt to bridge LLM values and their safety, revealing the predictive power of different value systems and the impacts of various values on LLM safety. Through interdisciplinary efforts, we aim to leverage AI for next-generation psychometrics and psychometrics for value-aligned AI.
论文检索
输入标题、作者或关键词，从 2,072 篇学术成果中精准定位
Multimodal Large Language Models (MLLMs) excel in solving text-based mathematical problems, but they struggle with mathematical diagrams since they are primarily trained on natural scene images. For humans, visual aids generally enhance problem-solving, but MLLMs perform worse as information shifts from textual to visual modality. This decline is mainly due to their shortcomings in aligning images and text. To tackle aforementioned challenges, we propose Math-PUMA, a methodology focused on Progressive Upward Multimodal Alignment. This approach is designed to improve the mathematical reasoning skills of MLLMs through a three-stage training process, with the second stage being the critical alignment stage. We first enhance the language model's mathematical reasoning capabilities with extensive set of textual mathematical problems. We then construct a multimodal dataset with varying degrees of textual and visual information, creating data pairs by presenting each problem in at least two forms. By leveraging the Kullback-Leibler (KL) divergence of next-token prediction distributions to align visual and textual modalities, consistent problem-solving abilities are ensured. Finally, we utilize multimodal instruction tuning for MLLMs with high-quality multimodal data. Experimental results on multiple mathematical reasoning benchmarks demonstrate that the MLLMs trained with Math-PUMA surpass most open-source MLLMs. Our approach effectively narrows the performance gap for problems presented in different modalities.
We introduce PokerBench - a benchmark for evaluating the poker-playing abilities of large language models (LLMs). As LLMs excel in traditional NLP tasks, their application to complex, strategic games like poker poses a new challenge. Poker, an incomplete information game, demands a multitude of skills such as mathematics, reasoning, planning, strategy, and a deep understanding of game theory and human psychology. This makes Poker the ideal next frontier for large language models. PokerBench consists of a comprehensive compilation of 11,000 most important scenarios, split between pre-flop and post-flop play, developed in collaboration with trained poker players. We evaluate prominent models including GPT-4, ChatGPT 3.5, and various Llama and Gemma series models, finding that all state-of-the-art LLMs underperform in playing optimal poker. However, after fine-tuning, these models show marked improvements. We validate PokerBench by having models with different scores compete with each other, demonstrating that higher scores on PokerBench leads to higher win rates in actual poker games. Through gameplay between our fine-tuned model and GPT-4, we also identify limitations of simple supervised fine-tuning for learning optimal playing strategy, suggesting the need for more advanced methodologies for effectively training language models to excel in games. PokerBench thus presents a unique benchmark for a quick and reliable evaluation of the poker-playing ability of LLMs as well as a comprehensive benchmark to study the progress of LLMs in complex game-playing scenarios.
Large language model (LLM) decoding involves generating a sequence of tokens based on a given context, where each token is predicted one at a time using the model's learned probabilities. The typical autoregressive decoding method requires a separate forward pass through the model for each token generated, which is computationally inefficient and poses challenges for deploying LLMs in latency-sensitive scenarios. The main limitations of current decoding methods stem from their inefficiencies and resource demands. Existing approaches either necessitate fine-tuning smaller models, which is resource-intensive, or relying on fixed retrieval schemes to construct drafts for the next tokens, which lack adaptability and fail to generalize across different models and contexts. To address these issues, we introduce a novel methodology called Adaptix, which accelerates LLM decoding without requiring fine-tuning. Our approach involves an adaptive draft-verification process that evolves over time to improve efficiency. We utilize a tri-gram matrix-based LLM representation to dynamically approximate the output distribution of the LLM, allowing the model to adjust to changing token probabilities during the decoding process. Additionally, we implement a draft construction mechanism that effectively balances exploration and exploitation, ensuring that the drafts generated are both diverse and close to the true output distribution of the LLM. The importance of this design lies in its ability to optimize the draft distribution adaptively, leading to faster and more accurate decoding. Through extensive experiments on various benchmark datasets and LLM architectures, we demonstrate that Adaptix significantly accelerates the decoding process while maintaining high accuracy, making it suitable for deployment in a wide range of practical applications.
The autoregressive decoding paradigm endows large language models (LLMs) with superior language generation capabilities; however, its step-by-step decoding process inherently limits decoding speed. To mitigate these constraints, the prevalent “draft and validation” strategy enables parallel validation of candidate drafts, allowing LLMs to decode multiple tokens simultaneously during one model forward propagation. However, existing methodologies for obtaining drafts often incur additional overhead in communication or training process, or statistical biases from the corpus. To this end, we propose an innovative draft generation and maintenance approach that leverages the capabilities of LLM itself. Specifically, we extend the autoregressive decoding paradigm to a multi-branch drafting procedure, which can efficiently generate draft sequences without any additional models or training process, while preserving the quality of the generated content by maintaining LLM parameters. Experiments across various open-source benchmarks show that our method generates 2.0 to 3.2 tokens per forward step and achieves around 2 times improvement of end-to-end throughput compared to the autoregressive decoding strategy.
This work presents a novel systematic methodology to analyse the capabilities and limitations of Large Language Models (LLMs) with feedback from a formal inference engine, on logic theory induction. The analysis is complexity-graded w.r.t. rule dependency structure, allowing quantification of specific inference challenges on LLM performance. Integrating LLMs with formal methods is a promising frontier in the Natural Language Processing field, as an important avenue for improving model inference control and explainability. In particular, inductive learning over complex sets of facts and rules, poses unique challenges for current autoregressive models, as they lack explicit symbolic grounding. While they can be complemented by formal systems, the properties delivered by LLMs regarding inductive learning, are not well understood and quantified. Empirical results indicate that the largest LLMs can achieve competitive results against a SOTA Inductive Logic Programming (ILP) system baseline, but also that tracking long predicate relationship chains is a more difficult obstacle than theory complexity for LLMs.
The rapid development of large language models (LLMs) has significantly advanced code completion capabilities, giving rise to a new generation of LLM-based Code Completion Tools (LCCTs). Unlike general-purpose LLMs, these tools possess unique workflows, integrating multiple information sources as input and prioritizing code suggestions over natural language interaction, which introduces distinct security challenges. Additionally, LCCTs often rely on proprietary code datasets for training, raising concerns about the potential exposure of sensitive data. This paper exploits these distinct characteristics of LCCTs to develop targeted attack methodologies on two critical security risks: jailbreaking and training data extraction attacks. Our experimental results expose significant vulnerabilities within LCCTs, including a 99.4% success rate in jailbreaking attacks on GitHub Copilot and a 46.3% success rate on Amazon Q. Furthermore, We successfully extracted sensitive user data from GitHub Copilot, including 54 real email addresses and 314 physical addresses associated with GitHub usernames. Our study also demonstrates that these code-based attack methods are effective against general-purpose LLMs, highlighting a broader security misalignment in the handling of code by modern LLMs. These findings underscore critical security challenges associated with LCCTs and suggest essential directions for strengthening their security frameworks.
SrSv: Integrating Sequential Rollouts with Sequential Value Estimation for Multi-agent Reinforcement Learning
PDF ↗Although multi-agent reinforcement learning (MARL) has shown its success across diverse domains, extending its application to large-scale real-world systems still faces significant challenges. Primarily, the high complexity of real-world environments exacerbates the credit assignment problem, substantially reducing training efficiency. Moreover, the variability of agent populations in large-scale scenarios necessitates scalable decision-making mechanisms. To address these challenges, we propose a novel framework: Sequential rollout with Sequential value estimation (SrSv). This framework aims to capture agent interdependence and provide a scalable solution for cooperative MARL. Specifically, SrSv leverages the autoregressive property of the Transformer model to handle varying populations through sequential action rollout. Furthermore, to capture the interdependence of policy distributions and value functions among multiple agents, we introduce an innovative sequential value estimation methodology and integrates the value approximation into an attention-based sequential model. We evaluate SrSv on three benchmarks: Multi-Agent MuJoCo, StarCraft Multi-Agent Challenge, and DubinsCars. Experimental results demonstrate that SrSv significantly outperforms baseline methods in terms of training efficiency without compromising convergence performance. Moreover, when implemented in a large-scale DubinsCar system with 1,024 agents, our framework surpasses existing benchmarks, highlighting the excellent scalability of SrSv.
Advancements in chip design and manufacturing have enabled the processing of complex tasks such as deep learning and natural language processing, paving the way for the development of artificial general intelligence (AGI). AI, on the other hand, can be leveraged to innovate and streamline semiconductor technology from planning and implementation to manufacturing. In this paper, we present Intelligent OPC Engineer Assistant, an AI/LLM-powered methodology designed to solve the core manufacturing-aware optimization problem known as Optical Proximity Correction (OPC). The methodology involves a reinforcement learning-based OPC recipe search and a customized multi-modal agent system for recipe summarization. Experiments demonstrate that our methodology can efficiently build OPC recipes on various chip designs with specially handled design topologies, a task that typically requires the full-time effort of OPC engineers with years of experience.
Backdoor attacks pose a significant threat during the model's training phase. Attackers craft pre-defined triggers to break deep neural networks, ensuring the model accurately classifies clean samples during inference yet erroneously classifies samples added with these triggers. Recent studies have shown that speaker recognition systems trained on large-scale data are susceptible to backdoor attacks. Existing attackers employ unnoticed ambient sounds as triggers. However, these sounds are not inherently part of the training samples themselves. In essence, triggers can be designed to maintain an intrinsic connection with the original speech to enhance stealthiness. Our paper presents a novel attack methodology named Speed Master, which undermines deep neural networks by manipulating the speed of speech samples. Specifically, we execute poison-only backdoor attacks using speed or tempo adjustment. Changes in speech rate have become a common occurrence, as seen on platforms that allow users to adjust playback speed. In real-world scenarios, people naturally adjust their speaking rate depending on the context. As a result, changes in a speaker’s speech rate are typically perceived as normal and are unlikely to raise suspicion. Furthermore, detecting such subtle adjustments becomes challenging for users without reference speech. Our comprehensive experiments demonstrate that Speed Master can achieve an ASR over 99% in the digital domain, with only a 0.6% poisoning rate. Additionally, we validate the feasibility of Speed Master in the real world and its resistance to typical defensive measures.
Diffusion-based models have been recently shown to be high-quality data generators. However, their performance severely degrades when training on non-stationary changing data distributions in an online manner, due to the catastrophic forgetting. In this paper, we propose enabling the diffusion model with a novel Dynamic Expansion Memory Unit (DEMU) methodology that adaptively creates new memory buffers, to be added to a memory system, in order to preserve information deemed critical for training the model. Having a selective memory unit is essential for training diffusion networks, which are expensive to train, especially when deployed in resource-constrained environments. A Maximum Mean Discrepancy (MMD) based expansion mechanism, that evaluates probabilistic distances between each of the previously defined memory buffers and the newly given data, and uses them as expansion signals, is employed for ensuring the diversity of information learning. We propose a new model expansion mechanism to automatically add new diffusion models as experts in a mixture system, which enhances the multi-domain image generation performance. Also a novel memory compaction approach is proposed to automatically remove statistically overlapping memory units, through a graph relationship evaluation, preventing the limitless expansion of DEMU. Comprehensive results show that the proposed approach performs better than the state-of-the-art.
The challenges tied to unstructured graph data are manifold, primarily falling into node, edge, and graph-level problem categories. Graph Neural Networks (GNNs) serve as effective tools to tackle these issues. However, individual tasks often demand distinct model architectures, and training these models typically requires abundant labeled data, a luxury often unavailable in practical settings. Recently, various "prompt tuning" methodologies have emerged to empower GNNs to adapt to multi-task learning with limited labels. The crux of these methods lies in bridging the gap between pre-training tasks and downstream objectives. Nonetheless, a prevalent oversight in existing studies is the homophily-centric nature of prompt tuning frameworks, disregarding scenarios characterized by high heterogeneity. To remedy this oversight, we introduce a novel prompting strategy named HeterGP tailored for highly heterophilic scenarios. Specifically, we present a dual-view approach to capture both homophilic and heterophilic information, along with a prompt graph design that encompasses token initialization and insertion patterns. Through extensive experiments conducted in a few-shot context encompassing node and graph classification tasks, our method showcases superior performance in highly heterophilic environments compared to state-of-the-art prompt tuning techniques.
Action advising endeavors to leverage supplementary guidance from expert teachers to alleviate the issue of sampling inefficiency in Deep Reinforcement Learning (DRL). Previous agent-specific action advising methods are hindered by imperfections in the agent itself, while agent-agnostic approaches exhibit limited adaptability to the learning agent. In this study, we propose a novel framework called Agent-Aware trAining yet Agent-Agnostic Action Advising (A7) to strike a balance between the two. The underlying concept of A7 revolves around utilizing the similarity of state features as an indicator for soliciting advice. However, unlike prior methodologies, the measurement of state feature similarity is performed by neither the error-prone learning agent nor the agent-agnostic advisor. Instead, we employ a proxy model to extract state features that are both discriminative (adaptive to the agent) and generally applicable (robust to agent noise). Furthermore, we utilize behavior cloning to train a model for reusing advice and introduce an intrinsic reward for the advised samples to incentivize the utilization of expert guidance. Experiments are conducted on the GridWorld, LunarLander, and six prominent scenarios from Atari games. The results demonstrate that A7 significantly accelerates the learning process and surpasses existing methods (both agent- specific and agent-agnostic) by a substantial margin. Our code will be made publicly available.
Open set anomaly detection (OSAD) is a crucial task that aims to identify abnormal patterns or behaviors in data sets, especially when the anomalies observed during training do not represent all possible classes of anomalies. The recent advances in quantum computing in handling complex data structures and improving machine learning models herald a paradigm shift in anomaly detection methodologies. This study proposes a Quantum Scoring Module (Qsco), embedding quantum variational circuits into neural networks to enhance the model's processing capabilities in handling uncertainty and unlabeled data. Extensive experiments conducted across eight real-world anomaly detection datasets demonstrate our model's superior performance in detecting anomalies across varied settings and reveal that integrating quantum simulators does not result in prohibitive time complexities. At the same time, the experimental results under different noise models also prove that Qsco is a noise-resilient algorithm. Our study validates the feasibility of quantum-enhanced anomaly detection methods in practical applications.
In recent years, employing Shapley values to compute feature importance has gained considerable attention. Calculating these values inherently necessitates managing an exponential number of parameters—a challenge commonly mitigated through an additivity assumption coupled with linear regression. This paper proposes a novel approach by modeling supervised learning as a multilinear game, incorporating both direct and interaction effects to establish the requisite values for Shapley value computation. To efficiently handle the exponentially increasing parameters intrinsic to multilinear games, we introduce a support vector machine (SVM)-based method for parameter estimation, its complexity is predominantly contingent on the number of samples due to the implementation of a dual SVM formulation. Additionally, we unveil an optimized dynamic programming algorithm capable of directly computing the Shapley value and interaction index from the dual SVM. Our proposed methodology is versatile and we demonstrate that it can be applied to local explanation and feature selection. Experiments underscore the competitive efficacy of our proposed methods in terms of feature selection and explanation.
Unlocking the Game: Estimating Games in Möbius Representation for Explanation and High-Order Interaction Detection
PDF ↗Shapley value-based explanations are widely utilized to demystify predictions made by opaque models. Approaches to estimating Shapley values often approximate explanation games as inessential and estimate the Shapley value directly as feature attribution with a limited capacity to quantify feature interactions. This paper introduces a new approach for calculating Shapley values that relaxes the assumption of inessential games and is proven to provide additive feature attribution. The initial formulation of the proposed approach includes the estimation of game values in their Möbius representation with exponentially many parameters, but we put forward a polynomial-time algorithm designed to manage the game's numerous values and achieve an efficient linear-time computation of the Shapley value. Moreover, this formulation uniquely enables identifying only the significant high-order feature interactions amidst a potentially exponential set. Through experiments, we demonstrate the robust performance of our methodology in game estimation and in providing explanations for multiple black-box models.
Molecular design inherently involves the optimization of multiple conflicting objectives, such as enhancing bio-activity and ensuring synthesizability. Evaluating these objectives often requires resource-intensive computations or physical experiments. Current molecular design methodologies typically approximate the Pareto set using a limited number of molecules. In this paper, we present an innovative approach, called Multi-Objective Molecular Design through Learning Latent Pareto Set (MLPS). MLPS initially utilizes an encoder-decoder model to seamlessly transform the discrete chemical space into a continuous latent space. We then employ local Bayesian optimization models to efficiently search for local optimal solutions (i.e., molecules) within predefined trust regions. Using surrogate objective values derived from these local models, we train a global Pareto set learning model to understand the mapping between direction vectors (called “preferences”) in the objective space and the entire Pareto set in the continuous latent space. Both the global Pareto set learning model and local Bayesian optimization models collaborate to discover high-quality solutions and adapt the trust regions dynamically. Our work is an effective endeavor towards learning the Pareto set for multi-objective molecular design, providing decision-makers with the capability to fine-tune their preferences and thoroughly explore the Pareto set. Experimental results demonstrate that MLPS achieves state-of-the-art performance across various multi-objective scenarios, encompassing diverse objective types and varying numbers of objectives. The effectiveness of MLPS was further validated through real-world challenges in discovering antifungal peptides with low toxicity and high activity.
Importance sampling is a rare event simulation technique used in Monte Carlo simulations to bias the sampling distribution towards the rare event of interest. By assigning appropriate weights to sampled points, importance sampling allows for more efficient estimation of rare events or tails of distributions. However, importance sampling can fail when the proposal distribution does not effectively cover the target distribution. In this work, we propose a method for more efficient sampling by updating the proposal distribution in the latent space of a normalizing flow. Normalizing flows learn an invertible mapping from a target distribution to a simpler latent distribution. The latent space can be more easily explored during the search for a proposal distribution, and samples from the proposal distribution are recovered in the space of the target distribution via the invertible mapping. We empirically validate our methodology on simulated robotics applications such as autonomous racing and aircraft ground collision avoidance.
Token compression techniques, such as token merging and pruning, are essential for alleviating the substantial computational burden caused by the proliferation of tokens within attention mechanisms. However, current methods often rely on token-to-token distances or similarity metrics to evaluate token importance, which is inadequate in the context of modern promptable designs and frameworks that are gaining prominence. To address this limitation, we introduce a novel and effective merging strategy called “Multimodal Promptable Token Merging” (MPTM). The proposed method leverages a multimodal, prompt-centric methodology, assessing the proximity between tokens of each input modality and the multimodal prompt to efficiently eliminate redundant tokens while preserving those rich in information. Extensive experiments demonstrate that MPTM significantly reduces computational costs without compromising essential information in generative image tasks. When integrated into diffusion-based detection architectures, MPTM outperforms existing state-of-the-art methods by 2.3% in object detection tasks. Additionally, when applied to multimodal diffusion models, MPTM maintains high-quality output while achieving a 2.9-fold increase in throughput, highlighting its versatility.
Auto-regressive models have made significant progress in the realm of text-to-image synthesis, yet devising an appropriate model architecture and training strategy to achieve a satisfactory level remains an important avenue of exploration. In this work, we introduce MARS, a novel framework for T2I generation that incorporates a specially designed Semantic Vision-Language Integration Expert (SemVIE). This innovative component integrates pre-trained LLMs by independently processing linguistic and visual information—freezing the textual component while fine-tuning the visual component. This methodology preserves the NLP capabilities of LLMs while imbuing them with exceptional visual understanding. Building upon the powerful base of the pre-trained Qwen-7B, MARS stands out with its bilingual generative capabilities corresponding to both English and Chinese language prompts and the capacity for joint image and text generation. The flexibility of this framework lends itself to migration towards any-to-any task adaptability. Furthermore, MARS employs a multi-stage training strategy that first establishes robust image-text alignment through complementary bidirectional tasks and subsequently concentrates on refining the T2I generation process, significantly augmenting text-image synchrony and the granularity of image details. Notably, MARS requires only 9% of the GPU days needed by SD1.5, yet it achieves remarkable results across a variety of benchmarks, illustrating the training efficiency and the potential for swift deployment in various applications.