While Retrieval-Augmented Generation (RAG) has emerged as an effective approach for addressing the knowledge outdating problem in Large Language Models (LLMs), it faces a critical challenge: the prevalence of outdated information in knowledge bases. Current research primarily focuses on incorporating up-to-date information, yet the impact of outdated information coexisting in retrieval sources remains inadequately addressed. To bridge this gap, we introduce HoH, the first benchmark specifically designed to evaluate the impact of outdated information on RAG. Our benchmark leverages token-level diff algorithms combined with LLM pipelines to efficiently create a large-scale QA dataset that accurately captures temporal knowledge evolution in real-world facts. Through comprehensive experiments, we reveal that outdated information significantly degrades RAG performance in two critical ways: (1) it substantially reduces response accuracy by distracting models from correct information, and (2) it can mislead models into generating potentially harmful outputs, even when current information is available. Current RAG approaches struggle with both retrieval and generation aspects when handling outdated information. These findings highlight the urgent need for innovative solutions to address the temporal challenges in RAG.

Fine-tuning of Large Language Models (LLMs) for downstream tasks, performed on domain-specific data has shown significant promise. However, commercial use of such LLMs is limited by the high computational cost required for their deployment at scale. On the other hand, small Language Models (LMs) are much more cost effective but have subpar performance in a similar setup. This paper presents our approach to finetuning a small LM, that reaches high accuracy in multiple choice question answering task. We achieve this by improving data quality at each stage of the LM training pipeline. In particular, we start with data structuring resulting in extraction of compact, semantically meaningful text chunks used by a retriever. This allows more efficient knowledge digestion by the LM. Further, we improve the retrieved context by training a lightweight Chunk Re-Ranker (CRR) that generates more accurate relative relevance chunk scores. Finally, we improve the model generalization ability by merging the models fine-tuned with different parameters on different data subsets. We present detailed procedure descriptions, and corresponding experimental findings that show the improvements of each one of the proposed techniques.

Intelligence is a crucial trait for species to find solutions within a limited number of trial-and-error attempts. Building on this idea, we introduce Survival Game as a framework to evaluate intelligence based on the number of failed attempts in a trial-and-error process. Fewer failures indicate higher intelligence. When the expectation and variance of failure counts are both finite, it signals the ability to consistently find solutions to new challenges, which we define as the Autonomous Level of intelligence. Using Survival Game, we comprehensively evaluate existing AI systems. Our results show that while AI systems achieve the Autonomous Level in simple tasks, they are still far from it in more complex tasks, such as vision, search, recommendation, and language. While scaling current AI technologies might help, this would come at an astronomical cost. Projections suggest that achieving the Autonomous Level for general tasks would require $10^{26}$ parameters. To put this into perspective, loading such a massive model requires so many H100 GPUs that their total value is $10^{7}$ times that of Apple Inc.'s market value. Even with Moore's Law, supporting such a parameter scale would take $70$ years. This staggering cost highlights the complexity of human tasks and the inadequacies of current AI technologies. To further investigate this phenomenon, we conduct a theoretical analysis of Survival Game and its experimental results. Our findings suggest that human tasks possess a criticality property. As a result, Autonomous Level requires a deep understanding of the task's underlying mechanisms. Current AI systems, however, do not fully grasp these mechanisms and instead rely on superficial mimicry, making it difficult for them to reach an autonomous level. We believe Survival Game can not only guide the future development of AI but also offer profound insights into human intelligence.

Recent advancements in Large Language Models (LLMs) have led to the development of intelligent LLM-based agents capable of interacting with graphical user interfaces (GUIs). These agents demonstrate strong reasoning and adaptability, enabling them to perform complex tasks that traditionally required predefined rules. However, the reliance on step-by-step reasoning in LLM-based agents often results in inefficiencies, particularly for routine tasks. In contrast, traditional rule-based systems excel in efficiency but lack the intelligence and flexibility to adapt to novel scenarios. To address this challenge, we propose a novel evolutionary framework for GUI agents that enhances operational efficiency while retaining intelligence and flexibility. Our approach incorporates a memory mechanism that records the agent's task execution history. By analyzing this history, the agent identifies repetitive action sequences and evolves high-level actions that act as shortcuts, replacing these low-level operations and improving efficiency. This allows the agent to focus on tasks requiring more complex reasoning, while simplifying routine actions. Experimental results on multiple benchmark tasks demonstrate that our approach significantly outperforms existing methods in both efficiency and accuracy. The code will be open-sourced to support further research.

Large Language Models as Natural Selector for Embodied Soft Robot Design Changhe Chen Xiaohao Xu Xiangdong Wang Xiaonan Huang Abstract -- Designing soft robots is a complex and iterative process that demands cross-disciplinary expertise in materials science, mechanics, and control, often relying on intuition and extensive experimentation. While Large Language Models (LLMs) have demonstrated impressive reasoning abilities, their capacity to learn and apply embodied design principles--crucial for creating functional robotic systems--remains largely unexplored. This paper introduces RoboCrafter-QA, a novel benchmark to evaluate whether LLMs can learn representations of soft robot designs that effectively bridge the gap between high-level task descriptions and low-level morphological and material choices. RoboCrafter-QA leverages the EvoGym simulator to generate a diverse set of soft robot design challenges, spanning robotic locomotion, manipulation, and balancing tasks. Our experiments with state-of-the-art multi-modal LLMs reveal that while these models exhibit promising capabilities in learning design representations, they struggle with fine-grained distinctions between designs with subtle performance differences. We further demonstrate the practical utility of LLMs for robot design initialization. Our code and benchmark will be available to encourage the community to foster this exciting research direction 1 .

Text-to-SQL, the task of translating natural language questions into SQL queries, plays a crucial role in enabling non-experts to interact with databases. While recent advancements in large language models (LLMs) have significantly enhanced text-to-SQL performance, existing approaches face notable limitations in real-world text-to-SQL applications. Prompting-based methods often depend on closed-source LLMs, which are expensive, raise privacy concerns, and lack customization. Fine-tuning-based methods, on the other hand, suffer from poor generalizability due to the limited coverage of publicly available training data. To overcome these challenges, we propose a novel and scalable text-to-SQL data synthesis framework for automatically synthesizing large-scale, high-quality, and diverse datasets without extensive human intervention. Using this framework, we introduce SynSQL-2.5M, the first million-scale text-to-SQL dataset, containing 2.5 million samples spanning over 16,000 synthetic databases. Each sample includes a database, SQL query, natural language question, and chain-of-thought (CoT) solution. Leveraging SynSQL-2.5M, we develop OmniSQL, a powerful open-source text-to-SQL model available in three sizes: 7B, 14B, and 32B. Extensive evaluations across nine datasets demonstrate that OmniSQL achieves state-of-the-art performance, matching or surpassing leading closed-source and open-source LLMs, including GPT-4o and DeepSeek-V3, despite its smaller size. We release all code, datasets, and models to support further research.

The advancement of Connected and Automated Vehicles (CAVs) and Vehicle-to-Everything (V2X) offers significant potential for enhancing transportation safety, mobility, and sustainability. However, the integration and analysis of the diverse and voluminous V2X data, including Basic Safety Messages (BSMs) and Signal Phase and Timing (SPaT) data, present substantial challenges, especially on Connected Vehicle Corridors. These challenges include managing large data volumes, ensuring real-time data integration, and understanding complex traffic scenarios. Although these projects have developed an advanced CAV data pipeline that enables real-time communication between vehicles, infrastructure, and other road users for managing connected vehicle and roadside unit (RSU) data, significant hurdles in data comprehension and real-time scenario analysis and reasoning persist. To address these issues, we introduce the V2X-LLM framework, a novel enhancement to the existing CV data pipeline. V2X-LLM leverages Large Language Models (LLMs) to improve the understanding and real-time analysis of V2X data. The framework includes four key tasks: Scenario Explanation, offering detailed narratives of traffic conditions; V2X Data Description, detailing vehicle and infrastructure statuses; State Prediction, forecasting future traffic states; and Navigation Advisory, providing optimized routing instructions. By integrating LLM-driven reasoning with V2X data within the data pipeline, the V2X-LLM framework offers real-time feedback and decision support for traffic management. This integration enhances the accuracy of traffic analysis, safety, and traffic optimization. Demonstrations in a real-world urban corridor highlight the framework's potential to advance intelligent transportation systems.

Vision-Language Models (VLMs) excel in integrating visual and textual information for vision-centric tasks, but their handling of inconsistencies between modalities is underexplored. We investigate VLMs' modality preferences when faced with visual data and varied textual inputs in vision-centered settings. By introducing textual variations to four vision-centric tasks and evaluating ten Vision-Language Models (VLMs), we discover a \emph{``blind faith in text''} phenomenon: VLMs disproportionately trust textual data over visual data when inconsistencies arise, leading to significant performance drops under corrupted text and raising safety concerns. We analyze factors influencing this text bias, including instruction prompts, language model size, text relevance, token order, and the interplay between visual and textual certainty. While certain factors, such as scaling up the language model size, slightly mitigate text bias, others like token order can exacerbate it due to positional biases inherited from language models. To address this issue, we explore supervised fine-tuning with text augmentation and demonstrate its effectiveness in reducing text bias. Additionally, we provide a theoretical analysis suggesting that the blind faith in text phenomenon may stem from an imbalance of pure text and multi-modal data during training. Our findings highlight the need for balanced training and careful consideration of modality interactions in VLMs to enhance their robustness and reliability in handling multi-modal data inconsistencies.

-- We present the first framework demonstrating zero-shot sim-to-real transfer of visual control policies learned in a Neural Radiance Field (NeRF) environment for quadrotors to fly through racing gates. Robust transfer from simulation to real flight poses a major challenge, as standard simulators often lack sufficient visual fidelity. T o address this, we construct a photorealistic simulation environment of quadrotor racing tracks, called FalconGym, which provides effectively unlimited synthetic images for training. Within FalconGym, we develop a pipelined approach for crossing gates that combines (i) a Neural Pose Estimator (NPE) coupled with a Kalman filter to reliably infer quadrotor poses from single-frame RGB images and IMU data, and (ii) a self-attention-based multi-modal controller that adaptively integrates visual features and pose estimation. This multi-modal design compensates for perception noise and intermittent gate visibility. We train this controller purely in FalconGym with imitation learning and deploy the resulting policy to real hardware with no additional fine-tuning. Simulation experiments on three distinct tracks (circle, U-turn and figure-8) demonstrate that our controller outperforms a vision-only state-of-the-art baseline in both success rate and gate-crossing accuracy. In 30 live hardware flights spanning three tracks and 120 gates, our controller achieves a 95.8% success rate and an average error of just 10 cm when flying through 38 cm-radius gates.

Large Language Model (LLM) agents have demonstrated remarkable generalization capabilities across multi-domain tasks. Existing agent tuning approaches typically employ supervised finetuning on entire expert trajectories. However, behavior-cloning of full trajectories can introduce expert bias and weaken generalization to states not covered by the expert data. Additionally, critical steps, such as planning, complex reasoning for intermediate subtasks, and strategic decision-making, are essential to success in agent tasks, so learning these steps is the key to improving LLM agents. For more effective and efficient agent tuning, we propose ATLaS that identifies the critical steps in expert trajectories and finetunes LLMs solely on these steps with reduced costs. By steering the training's focus to a few critical steps, our method mitigates the risk of overfitting entire trajectories and promotes generalization across different environments and tasks. In extensive experiments, an LLM finetuned on only 30% critical steps selected by ATLaS outperforms the LLM finetuned on all steps and recent open-source LLM agents. ATLaS maintains and improves base LLM skills as generalist agents interacting with diverse environments.