SEA-LION: Southeast Asian Languages in One Network

Recently, Large Language Models (LLMs) have dominated much of the artificial intelligence scene with their ability to process and generate natural languages. However, the majority of LLM research and development remains English-centric, leaving low-resource languages such as those in the Southeast Asian (SEA) region under-represented. To address this representation gap, we introduce Llama-SEA-LION-v3-8B-IT and Gemma-SEA-LION-v3-9B-IT, two cutting-edge multilingual LLMs designed for SEA languages. The SEA-LION family of LLMs supports 11 SEA languages, namely English, Chinese, Indonesian, Vietnamese, Malay, Thai, Burmese, Lao, Filipino, Tamil, and Khmer. Our work leverages large-scale multilingual continued pre-training with a comprehensive post-training regime involving multiple stages of instruction fine-tuning, alignment, and model merging. Evaluation results on multilingual benchmarks indicate that our models achieve state-of-the-art performance across LLMs supporting SEA languages. We open-source the models to benefit the wider SEA community.

ResearchBench: Benchmarking LLMs in Scientific Discovery via Inspiration-Based Task Decomposition

Large language models (LLMs) have demonstrated potential in assisting scientific research, yet their ability to discover high-quality research hypotheses remains unexamined due to the lack of a dedicated benchmark. To address this gap, we introduce the first large-scale benchmark for evaluating LLMs with a near-sufficient set of sub-tasks of scientific discovery: inspiration retrieval, hypothesis composition, and hypothesis ranking. We develop an automated framework that extracts critical components - research questions, background surveys, inspirations, and hypotheses - from scientific papers across 12 disciplines, with expert validation confirming its accuracy. To prevent data contamination, we focus exclusively on papers published in 2024, ensuring minimal overlap with LLM pretraining data. Our evaluation reveals that LLMs perform well in retrieving inspirations, an out-of-distribution task, suggesting their ability to surface novel knowledge associations. This positions LLMs as "research hypothesis mines", capable of facilitating automated scientific discovery by generating innovative hypotheses at scale with minimal human intervention.

Deep Cut-informed Graph Embedding and Clustering

Graph clustering aims to divide the graph into different clusters. The recently emerging deep graph clustering approaches are largely built on graph neural networks (GNN). However, GNN is designed for general graph encoding and there is a common issue of representation collapse in existing GNN-based deep graph clustering algorithms. We attribute two main reasons for such issue: (i) the inductive bias of GNN models: GNNs tend to generate similar representations for proximal nodes. Since graphs often contain a non-negligible amount of inter-cluster links, the bias results in error message passing and leads to biased clustering; (ii) the clustering guided loss function: most traditional approaches strive to make all samples closer to pre-learned cluster centers, which cause a degenerate solution assigning all data points to a single label thus make all samples and less discriminative. To address these challenges, we investigate graph clustering from a graph cut perspective and propose an innovative and non-GNN-based Deep Cut-informed Graph embedding and Clustering framework, namely DCGC. This framework includes two modules: (i) cut-informed graph encoding; (ii) self-supervised graph clustering via optimal transport. For the encoding module, we derive a cut-informed graph embedding objective to fuse graph structure and attributes by minimizing their joint normalized cut. For the clustering module, we utilize the optimal transport theory to obtain the clustering assignments, which can balance the guidance of proximity to the pre-learned cluster center. With the above two tailored designs, DCGC is more suitable for the graph clustering task, which can effectively alleviate the problem of representation collapse and achieve better performance. We conduct extensive experiments to demonstrate that our method is simple but effective compared with benchmarks.

Towards Robust ESG Analysis Against Greenwashing Risks: Aspect-Action Analysis with Cross-Category Generalization

Sustainability reports are key for evaluating companies' environmental, social and governance, ESG performance, but their content is increasingly obscured by greenwashing - sustainability claims that are misleading, exaggerated, and fabricated. Yet, existing NLP approaches for ESG analysis lack robustness against greenwashing risks, often extracting insights that reflect misleading or exaggerated sustainability claims rather than objective ESG performance. To bridge this gap, we introduce A3CG - Aspect-Action Analysis with Cross-Category Generalization, as a novel dataset to improve the robustness of ESG analysis amid the prevalence of greenwashing. By explicitly linking sustainability aspects with their associated actions, A3CG facilitates a more fine-grained and transparent evaluation of sustainability claims, ensuring that insights are grounded in verifiable actions rather than vague or misleading rhetoric. Additionally, A3CG emphasizes cross-category generalization. This ensures robust model performance in aspect-action analysis even when companies change their reports to selectively favor certain sustainability areas. Through experiments on A3CG, we analyze state-of-the-art supervised models and LLMs, uncovering their limitations and outlining key directions for future research.

ESGSenticNet: A Neurosymbolic Knowledge Base for Corporate Sustainability Analysis

Evaluating corporate sustainability performance is essential to drive sustainable business practices, amid the need for a more sustainable economy. However, this is hindered by the complexity and volume of corporate sustainability data (i.e. sustainability disclosures), not least by the effectiveness of the NLP tools used to analyse them. To this end, we identify three primary challenges - immateriality, complexity, and subjectivity, that exacerbate the difficulty of extracting insights from sustainability disclosures. To address these issues, we introduce ESGSenticNet, a publicly available knowledge base for sustainability analysis. ESGSenticNet is constructed from a neurosymbolic framework that integrates specialised concept parsing, GPT-4o inference, and semi-supervised label propagation, together with a hierarchical taxonomy. This approach culminates in a structured knowledge base of 44k knowledge triplets - ('halve carbon emission', supports, 'emissions control'), for effective sustainability analysis. Experiments indicate that ESGSenticNet, when deployed as a lexical method, more effectively captures relevant and actionable sustainability information from sustainability disclosures compared to state of the art baselines. Besides capturing a high number of unique ESG topic terms, ESGSenticNet outperforms baselines on the ESG relatedness and ESG action orientation of these terms by 26% and 31% respectively. These metrics describe the extent to which topic terms are related to ESG, and depict an action toward ESG. Moreover, when deployed as a lexical method, ESGSenticNet does not require any training, possessing a key advantage in its simplicity for non-technical stakeholders.

Gender Bias in Text-to-Video Generation Models: A case study of Sora

The advent of text-to-video generation models has revolutionized content creation as it produces high-quality videos from textual prompts. However, concerns regarding inherent biases in such models have prompted scrutiny, particularly regarding gender representation. Our study investigates the presence of gender bias in OpenAI's Sora, a state-of-the-art text-to-video generation model. We uncover significant evidence of bias by analyzing the generated videos from a diverse set of gender-neutral and stereotypical prompts. The results indicate that Sora disproportionately associates specific genders with stereotypical behaviors and professions, which reflects societal prejudices embedded in its training data.

Towards Rich Emotions in 3D Avatars: A Text-to-3D Avatar Generation Benchmark

Producing emotionally dynamic 3D facial avatars with text derived from spoken words (Emo3D) has been a pivotal research topic in 3D avatar generation. While progress has been made in general-purpose 3D avatar generation, the exploration of generating emotional 3D avatars remains scarce, primarily due to the complexities of identifying and rendering rich emotions from spoken words. This paper reexamines Emo3D generation and draws inspiration from human processes, breaking down Emo3D into two cascading steps: Text-to-3D Expression Mapping (T3DEM) and 3D Avatar Rendering (3DAR). T3DEM is the most crucial step in determining the quality of Emo3D generation and encompasses three key challenges: Expression Diversity, Emotion-Content Consistency, and Expression Fluidity. To address these challenges, we introduce a novel benchmark to advance research in Emo3D generation. First, we present EmoAva, a large-scale, high-quality dataset for T3DEM, comprising 15,000 text-to-3D expression mappings that characterize the aforementioned three challenges in Emo3D generation. Furthermore, we develop various metrics to effectively evaluate models against these identified challenges. Next, to effectively model the consistency, diversity, and fluidity of human expressions in the T3DEM step, we propose the Continuous Text-to-Expression Generator, which employs an autoregressive Conditional Variational Autoencoder for expression code generation, enhanced with Latent Temporal Attention and Expression-wise Attention mechanisms. Finally, to further enhance the 3DAR step on rendering higher-quality subtle expressions, we present the Globally-informed Gaussian Avatar (GiGA) model. GiGA incorporates a global information mechanism into 3D Gaussian representations, enabling the capture of subtle micro-expressions and seamless transitions between emotional states.

Towards Faithful Natural Language Explanations: A Study Using Activation Patching in Large Language Models

Large Language Models (LLMs) are capable of generating persuasive Natural Language Explanations (NLEs) to justify their answers. However, the faithfulness of these explanations should not be readily trusted at face value. Recent studies have proposed various methods to measure the faithfulness of NLEs, typically by inserting perturbations at the explanation or feature level. We argue that these approaches are neither comprehensive nor correctly designed according to the established definition of faithfulness. Moreover, we highlight the risks of grounding faithfulness findings on out-of-distribution samples. In this work, we leverage a causal mediation technique called activation patching, to measure the faithfulness of an explanation towards supporting the explained answer. Our proposed metric, Causal Faithfulness quantifies the consistency of causal attributions between explanations and the corresponding model outputs as the indicator of faithfulness. We experimented across models varying from 2B to 27B parameters and found that models that underwent alignment tuning tend to produce more faithful and plausible explanations. We find that Causal Faithfulness is a promising improvement over existing faithfulness tests by taking into account the model's internal computations and avoiding out of distribution concerns that could otherwise undermine the validity of faithfulness assessments. We release the code in \url{https://github.com/wj210/Causal-Faithfulness}

MOOSE-Chem: Large Language Models for Rediscovering Unseen Chemistry Scientific Hypotheses

Scientific discovery contributes largely to human society's prosperity, and recent progress shows that LLMs could potentially catalyze this process. However, it is still unclear whether LLMs can discover novel and valid hypotheses in chemistry. In this work, we investigate this central research question: Can LLMs automatically discover novel and valid chemistry research hypotheses given only a chemistry research background (consisting of a research question and/or a background survey), without limitation on the domain of the research question? After extensive discussions with chemistry experts, we propose an assumption that a majority of chemistry hypotheses can be resulted from a research background and several inspirations. With this key insight, we break the central question into three smaller fundamental questions. In brief, they are: (1) given a background question, whether LLMs can retrieve good inspirations; (2) with background and inspirations, whether LLMs can lead to hypothesis; and (3) whether LLMs can identify good hypotheses to rank them higher. To investigate these questions, we construct a benchmark consisting of 51 chemistry papers published in Nature, Science, or a similar level in 2024 (all papers are only available online since 2024). Every paper is divided by chemistry PhD students into three components: background, inspirations, and hypothesis. The goal is to rediscover the hypothesis, given only the background and a large randomly selected chemistry literature corpus consisting the ground truth inspiration papers, with LLMs trained with data up to 2023. We also develop an LLM-based multi-agent framework that leverages the assumption, consisting of three stages reflecting the three smaller questions. The proposed method can rediscover many hypotheses with very high similarity with the ground truth ones, covering the main innovations.

Negation Blindness in Large Language Models: Unveiling the NO Syndrome in Image Generation

Foundational Large Language Models (LLMs) have changed the way we perceive technology. They have been shown to excel in tasks ranging from poem writing and coding to essay generation and puzzle solving. With the incorporation of image generation capability, they have become more comprehensive and versatile AI tools. At the same time, researchers are striving to identify the limitations of these tools to improve them further. Currently identified flaws include hallucination, biases, and bypassing restricted commands to generate harmful content. In the present work, we have identified a fundamental limitation related to the image generation ability of LLMs, and termed it The NO Syndrome. This negation blindness refers to LLMs inability to correctly comprehend NO related natural language prompts to generate the desired images. Interestingly, all tested LLMs including GPT-4, Gemini, and Copilot were found to be suffering from this syndrome. To demonstrate the generalization of this limitation, we carried out simulation experiments and conducted entropy-based and benchmark statistical analysis tests on various LLMs in multiple languages, including English, Hindi, and French. We conclude that the NO syndrome is a significant flaw in current LLMs that needs to be addressed. A related finding of this study showed a consistent discrepancy between image and textual responses as a result of this NO syndrome. We posit that the introduction of a negation context-aware reinforcement learning based feedback loop between the LLMs textual response and generated image could help ensure the generated text is based on both the LLMs correct contextual understanding of the negation query and the generated visual output.