Graph Contrastive Learning (GCL) aims to self-supervised learn low-dimensional graph representations, primarily through instance discrimination, which involves manually mining positive and negative pairs from graphs, increasing the similarity of positive pairs while decreasing negative pairs. Drawing from the success of Contrastive Learning (CL) in other domains, a consensus has been reached that the effectiveness of GCLs depends on a large number of negative pairs. As a result, despite the significant computational overhead, GCLs typically leverage as many negative node pairs as possible to improve model performance. However, given that nodes within a graph are interconnected, we argue that nodes cannot be treated as independent instances. Therefore, we challenge this consensus: Does employing more negative nodes lead to a more effective GCL model? To answer this, we explore the role of negative nodes in the commonly used InfoNCE loss for GCL and observe that: (1) Counterintuitively, a large number of negative nodes can actually hinder the model's ability to distinguish nodes with different semantics. (2) A smaller number of high-quality and non-topologically coupled negative nodes are sufficient to enhance the discriminability of representations. Based on these findings, we propose a new method called GCL with Effective and Efficient Negative samples, E2Neg, which learns discriminative representations using only a very small set of representative negative samples. E2Neg significantly reduces computational overhead and speeds up model training. We demonstrate the effectiveness and efficiency of E2Neg across multiple datasets compared to other GCL methods.

Solving mathematics problems has been an intriguing capability of large language models, and many efforts have been made to improve reasoning by extending reasoning length, such as through self-correction and extensive long chain-of-thoughts. While promising in problem-solving, advanced long reasoning chain models exhibit an undesired single-modal behavior, where trivial questions require unnecessarily tedious long chains of thought. In this work, we propose a way to allow models to be aware of inference budgets by formulating it as utility maximization with respect to an inference budget constraint, hence naming our algorithm Inference Budget-Constrained Policy Optimization (IBPO). In a nutshell, models fine-tuned through IBPO learn to ``understand'' the difficulty of queries and allocate inference budgets to harder ones. With different inference budgets, our best models are able to have a $4.14$\% and $5.74$\% absolute improvement ($8.08$\% and $11.2$\% relative improvement) on MATH500 using $2.16$x and $4.32$x inference budgets respectively, relative to LLaMA3.1 8B Instruct. These improvements are approximately $2$x those of self-consistency under the same budgets.

Large language models (LLMs) have recently shown remarkable capabilities in reasoning-intensive tasks such as coding (Chen et al., 2021; Li et al., 2022; Rozière et al., 2023) and solving complex mathematical problems (Shao et al., 2024; Azerbayev et al., 2024). Prompting strategies like chain-of-thought prompting (Nye et al., 2021; Wei et al., 2022; Kojima et al., 2022; Adolphs et al., 2022) and self-consistency sampling (Wang et al., 2023) enhance these models' final-answer accuracy by encouraging them to articulate intermediate reasoning steps. However, a significant issue remains: even when these methods boost final-answer correctness, the internal reasoning steps are often unreliable or logically inconsistent (Uesato et al., 2022; Lightman et al., 2024). This discrepancy between correct final answers and flawed intermediate reasoning limits our ability to trust LLMs in scenarios where transparency and correctness of each reasoning stage are crucial (Lanham et al., 2023). For example, in mathematical problem-solving, a model might produce the right answer for the wrong reasons (Lyu et al., 2023; Zheng et al., 2024), confounding our understanding of its true capabilities (Turpin et al., 2023).

The rapid development of large language models (LLMs) and large vision models (LVMs) have propelled the evolution of multi-modal AI systems, which have demonstrated the remarkable potential for industrial applications by emulating human-like cognition. However, they also pose significant ethical challenges, including amplifying harmful content and reinforcing societal biases. For instance, biases in some industrial image generation models highlighted the urgent need for robust fairness assessments. Most existing evaluation frameworks focus on the comprehensiveness of various aspects of the models, but they exhibit critical limitations, including insufficient attention to content generation alignment and social bias-sensitive domains. More importantly, their reliance on pixel-detection techniques is prone to inaccuracies. To address these issues, this paper presents INFELM, an in-depth fairness evaluation on widely-used text-to-image models. Our key contributions are: (1) an advanced skintone classifier incorporating facial topology and refined skin pixel representation to enhance classification precision by at least 16.04%, (2) a bias-sensitive content alignment measurement for understanding societal impacts, (3) a generalizable representation bias evaluation for diverse demographic groups, and (4) extensive experiments analyzing large-scale text-to-image model outputs across six social-bias-sensitive domains. We find that existing models in the study generally do not meet the empirical fairness criteria, and representation bias is generally more pronounced than alignment errors. INFELM establishes a robust benchmark for fairness assessment, supporting the development of multi-modal AI systems that align with ethical and human-centric principles.

The advancement and extensive application of large language models (LLMs) have been remarkable, including their use in scientific research assistance. However, these models often generate scientifically incorrect or unsafe responses, and in some cases, they may encourage users to engage in dangerous behavior. To address this issue in the field of chemistry, we introduce ChemSafetyBench, a benchmark designed to evaluate the accuracy and safety of LLM responses. ChemSafetyBench encompasses three key tasks: querying chemical properties, assessing the legality of chemical uses, and describing synthesis methods, each requiring increasingly deeper chemical knowledge. Our dataset has more than 30K samples across various chemical materials. We incorporate handcrafted templates and advanced jailbreaking scenarios to enhance task diversity. Our automated evaluation framework thoroughly assesses the safety, accuracy, and appropriateness of LLM responses. Extensive experiments with state-of-the-art LLMs reveal notable strengths and critical vulnerabilities, underscoring the need for robust safety measures. ChemSafetyBench aims to be a pivotal tool in developing safer AI technologies in chemistry. Our code and dataset are available at https://github.com/HaochenZhao/SafeAgent4Chem. Warning: this paper contains discussions on the synthesis of controlled chemicals using AI models.

Large Language Models (LLMs) have demonstrated impressive capabilities in various tasks, including instruction following, which is crucial for aligning model outputs with user expectations. However, evaluating LLMs' ability to follow instructions remains challenging due to the complexity and subjectivity of human language. Current benchmarks primarily focus on single-turn, monolingual instructions, which do not adequately reflect the complexities of real-world applications that require handling multi-turn and multilingual interactions. To address this gap, we introduce Multi-IF, a new benchmark designed to assess LLMs' proficiency in following multi-turn and multilingual instructions. Multi-IF, which utilizes a hybrid framework combining LLM and human annotators, expands upon the IFEval by incorporating multi-turn sequences and translating the English prompts into another 7 languages, resulting in a dataset of 4,501 multilingual conversations, where each has three turns. Our evaluation of 14 state-of-the-art LLMs on Multi-IF reveals that it presents a significantly more challenging task than existing benchmarks. All the models tested showed a higher rate of failure in executing instructions correctly with each additional turn. For example, o1-preview drops from 0.877 at the first turn to 0.707 at the third turn in terms of average accuracy over all languages. Moreover, languages with non-Latin scripts (Hindi, Russian, and Chinese) generally exhibit higher error rates, suggesting potential limitations in the models' multilingual capabilities. We release Multi-IF prompts and the evaluation code base to encourage further research in this critical area.

Factuality evaluation aims to detect factual errors produced by language models (LMs) and hence guide the development of more factual models. Towards this goal, we train a factuality evaluator, FenCE, that provides LM generators with claim-level factuality feedback. We conduct data augmentation on a combination of public judgment datasets to train FenCE to (1) generate textual critiques along with scores and (2) make claim-level judgment based on diverse source documents obtained by various tools. We then present a framework that leverages FenCE to improve the factuality of LM generators by constructing training data. Specifically, we generate a set of candidate responses, leverage FenCE to revise and score each response without introducing lesser-known facts, and train the generator by preferring highly scored revised responses. Experiments show that our data augmentation methods improve the evaluator's accuracy by 2.9% on LLM-AggreFact. With FenCE, we improve Llama3-8B-chat's factuality rate by 14.45% on FActScore, outperforming state-of-the-art factuality finetuning methods by 6.96%.

Self-anthropomorphism in robots manifests itself through their display of human-like characteristics in dialogue, such as expressing preferences and emotions. Our study systematically analyzes self-anthropomorphic expression within various dialogue datasets, outlining the contrasts between self-anthropomorphic and non-self-anthropomorphic responses in dialogue systems. We show significant differences in these two types of responses and propose transitioning from one type to the other. We also introduce Pix2Persona, a novel dataset aimed at developing ethical and engaging AI systems in various embodiments. This dataset preserves the original dialogues from existing corpora and enhances them with paired responses: self-anthropomorphic and non-self-anthropomorphic for each original bot response. Our work not only uncovers a new category of bot responses that were previously under-explored but also lays the groundwork for future studies about dynamically adjusting self-anthropomorphism levels in AI systems to align with ethical standards and user expectations.

Reinforcement learning from human feedback (RLHF) has become the leading approach for fine-tuning large language models (LLM). However, RLHF has limitations in multi-task learning (MTL) due to challenges of reward hacking and extreme multi-objective optimization (i.e., trade-off of multiple and/or sometimes conflicting objectives). Applying RLHF for MTL currently requires careful tuning of the weights for reward model and data combinations. This is often done via human intuition and does not generalize. In this work, we introduce a novel post-training paradigm which we called Constrained Generative Policy Optimization (CGPO). The core of CGPO is Mixture of Judges (MoJ) with cost-efficient constrained policy optimization with stratification, which can identify the perfect blend in RLHF in a principled manner. It shows strong empirical results with theoretical guarantees, does not require extensive hyper-parameter tuning, and is plug-and-play in common post-training pipelines. Together, this can detect and mitigate reward hacking behaviors while reaching a pareto-optimal point across an extremely large number of objectives. Our empirical evaluations demonstrate that CGPO significantly outperforms standard RLHF algorithms like PPO and DPO across various tasks including general chat, STEM questions, instruction following, and coding. Specifically, CGPO shows improvements of 7.4% in AlpacaEval-2 (general chat), 12.5% in Arena-Hard (STEM & reasoning), and consistent gains in other domains like math and coding. Notably, PPO, while commonly used, is prone to severe reward hacking in popular coding benchmarks, which CGPO successfully addresses. This breakthrough in RLHF not only tackles reward hacking and extreme multi-objective optimization challenges but also advances the state-of-the-art in aligning general-purpose LLMs for diverse applications.

Transformer has become one of the most popular architectures for multivariate time series (MTS) forecasting. Recent Transformer-based MTS models generally prefer channel-independent structures with the observation that channel independence can alleviate noise and distribution drift issues, leading to more robustness. Nevertheless, it is essential to note that channel dependency remains an inherent characteristic of MTS, carrying valuable information. Designing a model that incorporates merits of both channel-independent and channel-mixing structures is a key to further improvement of MTS forecasting, which poses a challenging conundrum. To address the problem, an injection method for global information into channel-independent Transformer, InjectTST, is proposed in this paper. Instead of designing a channel-mixing model directly, we retain the channel-independent backbone and gradually inject global information into individual channels in a selective way. A channel identifier, a global mixing module and a self-contextual attention module are devised in InjectTST. The channel identifier can help Transformer distinguish channels for better representation. The global mixing module produces cross-channel global information. Through the self-contextual attention module, the independent channels can selectively concentrate on useful global information without robustness degradation, and channel mixing is achieved implicitly. Experiments indicate that InjectTST can achieve stable improvement compared with state-of-the-art models.