Large Language Models (LLMs) have demonstrated remarkable potential in advancing scientific knowledge and addressing complex challenges. In this work, we introduce OmniScience, a specialized large reasoning model for general science, developed through three key components: (1) domain adaptive pretraining on a carefully curated corpus of scientific literature, (2) instruction tuning on a specialized dataset to guide the model in following domain-specific tasks, and (3) reasoning-based knowledge distillation through fine-tuning to significantly enhance its ability to generate contextually relevant and logically sound responses. We demonstrate the versatility of OmniScience by developing a battery agent that efficiently ranks molecules as potential electrolyte solvents or additives. Comprehensive evaluations reveal that OmniScience is competitive with state-of-the-art large reasoning models on the GPQA Diamond and domain-specific battery benchmarks, while outperforming all public reasoning and non-reasoning models with similar parameter counts. We further demonstrate via ablation experiments that domain adaptive pretraining and reasoning-based knowledge distillation are critical to attain our performance levels, across benchmarks.

Large Language Models often generate factually incorrect but plausible outputs, known as hallucinations. We identify a more insidious phenomenon, LLM delusion, defined as high belief hallucinations, incorrect outputs with abnormally high confidence, making them harder to detect and mitigate. Unlike ordinary hallucinations, delusions persist with low uncertainty, posing significant challenges to model reliability. Through empirical analysis across different model families and sizes on several Question Answering tasks, we show that delusions are prevalent and distinct from hallucinations. LLMs exhibit lower honesty with delusions, which are harder to override via finetuning or self reflection. We link delusion formation with training dynamics and dataset noise and explore mitigation strategies such as retrieval augmented generation and multi agent debating to mitigate delusions. By systematically investigating the nature, prevalence, and mitigation of LLM delusions, our study provides insights into the underlying causes of this phenomenon and outlines future directions for improving model reliability.

Recent advancements in tool learning have enabled large language models (LLMs) to integrate external tools, enhancing their task performance by expanding their knowledge boundaries. However, relying on tools often introduces tradeoffs between performance, speed, and cost, with LLMs sometimes exhibiting overreliance and overconfidence in tool usage. This paper addresses the challenge of aligning LLMs with their knowledge boundaries to make more intelligent decisions about tool invocation. We propose a multi objective alignment framework that combines probabilistic knowledge boundary estimation with dynamic decision making, allowing LLMs to better assess when to invoke tools based on their confidence. Our framework includes two methods for knowledge boundary estimation, consistency based and absolute estimation, and two training strategies for integrating these estimates into the model decision making process. Experimental results on various tool invocation scenarios demonstrate the effectiveness of our framework, showing significant improvements in tool efficiency by reducing unnecessary tool usage.

A long-cherished vision of drones is to autonomously traverse through clutter to reach every corner of the world using onboard sensing and computation. In this paper, we combine onboard 3D lidar sensing and sim-to-real reinforcement learning (RL) to enable autonomous flight in cluttered environments. Compared to vision sensors, lidars appear to be more straightforward and accurate for geometric modeling of surroundings, which is one of the most important cues for successful obstacle avoidance. On the other hand, sim-to-real RL approach facilitates the realization of low-latency control, without the hierarchy of trajectory generation and tracking. We demonstrate that, with design choices of practical significance, we can effectively combine the advantages of 3D lidar sensing and RL to control a quadrotor through a low-level control interface at 50Hz. The key to successfully learn the policy in a lightweight way lies in a specialized surrogate of the lidar's raw point clouds, which simplifies learning while retaining a fine-grained perception to detect narrow free space and thin obstacles. Simulation statistics demonstrate the advantages of the proposed system over alternatives, such as performing easier maneuvers and higher success rates at different speed constraints. With lightweight simulation techniques, the policy trained in the simulator can control a physical quadrotor, where the system can dodge thin obstacles and safely traverse randomly distributed obstacles.

Zero-shot named entity recognition (NER) aims to develop entity recognition systems from unannotated text corpora. This task presents substantial challenges due to minimal human intervention. Recent work has adapted large language models (LLMs) for zero-shot NER by crafting specialized prompt templates. It advances model self-learning abilities by incorporating self-annotated demonstrations. However, two important challenges persist: (i) Correlations between contexts surrounding entities are overlooked, leading to wrong type predictions or entity omissions. (ii) The indiscriminate use of task demonstrations, retrieved through shallow similarity-based strategies, severely misleads LLMs during inference. In this paper, we introduce the cooperative multi-agent system (CMAS), a novel framework for zero-shot NER that uses the collective intelligence of multiple agents to address the challenges outlined above. CMAS has four main agents: (i) a self-annotator, (ii) a type-related feature (TRF) extractor, (iii) a demonstration discriminator, and (iv) an overall predictor. To explicitly capture correlations between contexts surrounding entities, CMAS reformulates NER into two subtasks: recognizing named entities and identifying entity type-related features within the target sentence. To enable controllable utilization of demonstrations, a demonstration discriminator is established to incorporate the self-reflection mechanism, automatically evaluating helpfulness scores for the target sentence. Experimental results show that CMAS significantly improves zero-shot NER performance across six benchmarks, including both domain-specific and general-domain scenarios. Furthermore, CMAS demonstrates its effectiveness in few-shot settings and with various LLM backbones.

Vision-and-Language Navigation (VLN) is an essential skill for embodied agents, allowing them to navigate in 3D environments following natural language instructions. High-performance navigation models require a large amount of training data, the high cost of manually annotating data has seriously hindered this field. Therefore, some previous methods translate trajectory videos into step-by-step instructions for expanding data, but such instructions do not match well with users' communication styles that briefly describe destinations or state specific needs. Moreover, local navigation trajectories overlook global context and high-level task planning. To address these issues, we propose NavRAG, a retrieval-augmented generation (RAG) framework that generates user demand instructions for VLN. NavRAG leverages LLM to build a hierarchical scene description tree for 3D scene understanding from global layout to local details, then simulates various user roles with specific demands to retrieve from the scene tree, generating diverse instructions with LLM. We annotate over 2 million navigation instructions across 861 scenes and evaluate the data quality and navigation performance of trained models.

Deep learning models can suffer from severe performance degradation when relying on spurious correlations between input features and labels, making the models perform well on training data but have poor prediction accuracy for minority groups. This problem arises especially when training data are limited or imbalanced. While most prior work focuses on learning invariant features (with consistent correlations to y), it overlooks the potential harm of spurious correlations between features. We hereby propose Elastic Representation (ElRep) to learn features by imposing Nuclear- and Frobenius-norm penalties on the representation from the last layer of a neural network. Similar to the elastic net, ElRep enjoys the benefits of learning important features without losing feature diversity. The proposed method is simple yet effective. It can be integrated into many deep learning approaches to mitigate spurious correlations and improve group robustness. Moreover, we theoretically show that ElRep has minimum negative impacts on in-distribution predictions. This is a remarkable advantage over approaches that prioritize minority groups at the cost of overall performance.

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve over 90\% accuracy on popular benchmarks like MMLU, limiting informed measurement of state-of-the-art LLM capabilities. In response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be the final closed-ended academic benchmark of its kind with broad subject coverage. HLE consists of 3,000 questions across dozens of subjects, including mathematics, humanities, and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable, but cannot be quickly answered via internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a significant gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai.

As the capabilities of code large language models (LLMs) continue to expand, their applications across diverse code intelligence domains are rapidly increasing. However, most existing datasets only evaluate limited application domains. To address this gap, we have developed a comprehensive code evaluation dataset FullStack Bench focusing on full-stack programming, which encompasses a wide range of application domains (e.g., basic programming, data analysis, software engineering, mathematics, and machine learning). Besides, to assess multilingual programming capabilities, in FullStack Bench, we design real-world instructions and corresponding unit test cases from 16 widely-used programming languages to reflect real-world usage scenarios rather than simple translations. Moreover, we also release an effective code sandbox execution tool (i.e., SandboxFusion) supporting various programming languages and packages to evaluate the performance of our FullStack Bench efficiently. Comprehensive experimental results on our FullStack Bench demonstrate the necessity and effectiveness of our FullStack Bench and SandboxFusion.

Comparing two (large) language models (LMs) side-by-side and pinpointing their prediction similarities and differences on the same set of inputs are crucial in many real-world scenarios, e.g., one can test if a licensed model was potentially plagiarized by another. Traditional analysis compares the LMs' outputs on some benchmark datasets, which only cover a limited number of inputs of designed perspectives for the intended applications. The benchmark datasets cannot prepare data to cover the test cases from unforeseen perspectives which can help us understand differences between models unbiasedly. In this paper, we propose a new model comparative analysis setting that considers a large input space where brute-force enumeration would be infeasible. The input space can be simply defined as all token sequences that a LM would produce low perplexity on -- we follow this definition in the paper as it would produce the most human-understandable inputs. We propose a novel framework \our that uses text generation by sampling and deweights the histogram of sampling statistics to estimate prediction differences between two LMs in this input space efficiently and unbiasedly. Our method achieves this by drawing and counting the inputs at each prediction difference value in negative log-likelihood. Experiments reveal for the first time the quantitative prediction differences between LMs in a large input space, potentially facilitating the model analysis for applications such as model plagiarism.