The attention operator remains a critical performance bottleneck in large language models (LLMs), particularly for long-context scenarios. While FlashAttention is the most widely used and effective GPU-aware acceleration algorithm, it must require time-consuming and hardware-specific manual implementation, limiting adaptability across GPU architectures. Existing LLMs have shown a lot of promise in code generation tasks, but struggle to generate high-performance attention code. The key challenge is it cannot comprehend the complex data flow and computation process of the attention operator and utilize low-level primitive to exploit GPU performance. To address the above challenge, we propose an LLM-friendly Thinking Language (LLM-TL) to help LLMs decouple the generation of high-level optimization logic and low-level implementation on GPU, and enhance LLMs' understanding of attention operator. Along with a 2-stage reasoning workflow, TL-Code generation and translation, the LLMs can automatically generate FlashAttention implementation on diverse GPUs, establishing a self-optimizing paradigm for generating high-performance attention operators in attention-centric algorithms. Verified on A100, RTX8000, and T4 GPUs, the performance of our methods significantly outshines that of vanilla LLMs, achieving a speed-up of up to 35.16x. Besides, our method not only surpasses human-optimized libraries (cuDNN and official library) in most scenarios but also extends support to unsupported hardware and data types, reducing development time from months to minutes compared with human experts.

The emergence of long-context large language models (LLMs) has enabled the use of hundreds, or even thousands, of demonstrations for in-context learning (ICL) - a previously impractical regime. This paper investigates whether traditional ICL selection strategies, which balance the similarity of ICL examples to the test input (using a text retriever) with diversity within the ICL set, remain effective when utilizing a large number of demonstrations. Our experiments demonstrate that, while longer contexts can accommodate more examples, simply increasing the number of demonstrations does not guarantee improved performance. Smart ICL selection remains crucial, even with thousands of demonstrations. To further enhance ICL in this setting, we introduce Refract ICL, a novel ICL selection algorithm specifically designed to focus LLM attention on challenging examples by strategically repeating them within the context and incorporating zero-shot predictions as error signals. Our results show that Refract ICL significantly improves the performance of extremely long-context models such as Gemini 1.5 Pro, particularly on tasks with a smaller number of output classes.

We present SheetMind, a modular multi-agent framework powered by large language models (LLMs) for spreadsheet automation via natural language instructions. The system comprises three specialized agents: a Manager Agent that decomposes complex user instructions into subtasks; an Action Agent that translates these into structured commands using a Backus Naur Form (BNF) grammar; and a Reflection Agent that validates alignment between generated actions and the user's original intent. Integrated into Google Sheets via a Workspace extension, SheetMind supports real-time interaction without requiring scripting or formula knowledge. Experiments on benchmark datasets demonstrate an 80 percent success rate on single step tasks and approximately 70 percent on multi step instructions, outperforming ablated and baseline variants. Our results highlight the effectiveness of multi agent decomposition and grammar based execution for bridging natural language and spreadsheet functionalities.

This paper investigates the influence of cognitive biases on Large Language Models (LLMs) outputs. Cognitive biases, such as confirmation and availability biases, can distort user inputs through prompts, potentially leading to unfaithful and misleading outputs from LLMs. Using a systematic framework, our study introduces various cognitive biases into prompts and assesses their impact on LLM accuracy across multiple benchmark datasets, including general and financial Q&A scenarios. The results demonstrate that even subtle biases can significantly alter LLM answer choices, highlighting a critical need for bias-aware prompt design and mitigation strategy. Additionally, our attention weight analysis highlights how these biases can alter the internal decision-making processes of LLMs, affecting the attention distribution in ways that are associated with output inaccuracies. This research has implications for Al developers and users in enhancing the robustness and reliability of Al applications in diverse domains.

Large Language Model (LLM) libraries have emerged as the foundational infrastructure powering today's AI revolution, serving as the backbone for LLM deployment, inference optimization, fine-tuning, and production serving across diverse applications. Despite their critical role in the LLM ecosystem, these libraries face frequent quality issues and bugs that threaten the reliability of AI systems built upon them. To address this knowledge gap, we present the first comprehensive empirical investigation into bug characteristics and testing practices in modern LLM libraries. We examine 313 bug-fixing commits extracted across two widely-adopted LLM libraries: HuggingFace Transformers and vLLM.Through rigorous manual analysis, we establish comprehensive taxonomies categorizing bug symptoms into 5 types and root causes into 14 distinct categories.Our primary discovery shows that API misuse has emerged as the predominant root cause (32.17%-48.19%), representing a notable transition from algorithm-focused defects in conventional deep learning frameworks toward interface-oriented problems. Additionally, we examine 7,748 test functions to identify 7 distinct test oracle categories employed in current testing approaches, with predefined expected outputs (such as specific tensors and text strings) being the most common strategy. Our assessment of existing testing effectiveness demonstrates that the majority of bugs escape detection due to inadequate test cases (41.73%), lack of test drivers (32.37%), and weak test oracles (25.90%). Drawing from these findings, we offer some recommendations for enhancing LLM library quality assurance.

Navigating the vast and rapidly growing body of scientific literature is a formidable challenge for aspiring researchers. Current approaches to supporting research idea generation often rely on generic large language models (LLMs). While LLMs are effective at aiding comprehension and summarization, they often fall short in guiding users toward practical research ideas due to their limitations. In this study, we present a novel structural framework for research ideation. Our framework, The Budget AI Researcher, uses retrieval-augmented generation (RAG) chains, vector databases, and topic-guided pairing to recombine concepts from hundreds of machine learning papers. The system ingests papers from nine major AI conferences, which collectively span the vast subfields of machine learning, and organizes them into a hierarchical topic tree. It uses the tree to identify distant topic pairs, generate novel research abstracts, and refine them through iterative self-evaluation against relevant literature and peer reviews, generating and refining abstracts that are both grounded in real-world research and demonstrably interesting. Experiments using LLM-based metrics indicate that our method significantly improves the concreteness of generated research ideas relative to standard prompting approaches. Human evaluations further demonstrate a substantial enhancement in the perceived interestingness of the outputs. By bridging the gap between academic data and creative generation, the Budget AI Researcher offers a practical, free tool for accelerating scientific discovery and lowering the barrier for aspiring researchers. Beyond research ideation, this approach inspires solutions to the broader challenge of generating personalized, context-aware outputs grounded in evolving real-world knowledge.

Tokyo 152 - 8552, Japan E - mail: kan.hatakeyama [ [ at ] ] weblab.t.u - tokyo.ac.jp Abstract This review explores the potential of foundation models to advanc e laboratory automation in the materials and chemical sciences. It emphasizes the dual roles of these models: cognitive functions for experimental planning and data analysis, and physical functions for hardware operations. While traditional laboratory automation has relied heavily on specialized, rigid systems, foundation models offer adaptability through their general - purpose intelligence and multimodal capabilities. Recent advancements have demonstrated the fea sibility of using large language models (LLMs) and multimodal robotic systems to handle complex and dynamic laboratory tasks. However, significant challenges remain, including precision manipulation of hardware, integration of multimodal data, and ensuring operational safety. Th is paper outlines a roadmap highlighting future directions, advocating for close interdisciplinary collaboration, benchmark establishment, and strategic human - AI integration to realize fully autonomous experimental laboratories. Keywords Laboratory Automation; Foundation Models; Robotics; Artificial Intelligence; Materials Science 1. Expectations for Foundation Models in Materials Laboratory Automation Laboratory automation, a technology aimed at automating experimental research, is expected to pave the way for a new research paradigm in materials science [1, 2, 3] . By rapidly and comprehensively executing numerous experiments, laboratory automation accelerates research, enhances reproducibility through precisely controlled robotic processes, and enables swift and distributed knowledge sharing among researchers worldwide [1] . This technology is anticipated to contribute significantly to the development of crucial devices and compounds, including catalyst s for energy and chemical conversions, environmentally friendly plastics, solar cells, secondary batteries, fuel cells, thermoelectric conversion modules, nuclear fusion reactors, quantum computers, and energy - efficient computing systems [1, 4, 5] . The success of next - generation laboratory automation depends not only o n experimental hardware but also o n the utilization of artificial intelligence (AI), especially foundation models. Foundation models represent a new AI paradigm encompassing large language models like GPT - 4 [6], multimodal models, and agent - related technologies. These foundation models and generative AI have begun to influenc e chemistry and materials science [7], giving rise to diverse applications including molecular and materials design [8, 9, 10], reaction pathway exploration [11], catalyst design [12], and even autonomous planning of chemical experiments [13] . Additionally, foundation models are being expanded to hardware control mechanisms, enabling natural language - driven robotic operations [14, 15] .

Chain-of-thought (CoT) prompting has been widely adopted to enhance the reasoning capabilities of large language models (LLMs). However, the effectiveness of CoT reasoning is inconsistent across tasks with different reasoning types. This work presents a novel perspective to understand CoT behavior through the lens of \textit{confirmation bias} in cognitive psychology. Specifically, we examine how model internal beliefs, approximated by direct question-answering probabilities, affect both reasoning generation ($Q \to R$) and reasoning-guided answer prediction ($QR \to A$) in CoT. By decomposing CoT into a two-stage process, we conduct a thorough correlation analysis in model beliefs, rationale attributes, and stage-wise performance. Our results provide strong evidence of confirmation bias in LLMs, such that model beliefs not only skew the reasoning process but also influence how rationales are utilized for answer prediction. Furthermore, the interplay between task vulnerability to confirmation bias and the strength of beliefs also provides explanations for CoT effectiveness across reasoning tasks and models. Overall, this study provides a valuable insight for the needs of better prompting strategies that mitigate confirmation bias to enhance reasoning performance. Code is available at \textit{https://github.com/yuewan2/biasedcot}.

We introduce TestCase-Eval, a new benchmark for systematic evaluation of LLMs in test-case generation. TestCase-Eval includes 500 algorithm problems and 100,000 human-crafted solutions from the Codeforces platform. It focuses on two pivotal tasks: (1) Fault Coverage, which measures how well LLM-generated test sets probe diverse input scenarios and cover a wide range of potential failure modes. (2) Fault Exposure, which evaluates whether LLMs can craft a tailored test input that reveals a specific incorrect code implementation. We provide a comprehensive assessment of 19 state-of-the-art open-source and proprietary LLMs on TestCase-Eval, offering insights into their strengths and limitations in generating effective test cases for algorithm problems.

Large Language Models (LLMs) have demonstrated remarkable capabilities across various domains. However, their potential to generate harmful responses has raised significant societal and regulatory concerns, especially when manipulated by adversarial techniques known as "jailbreak" attacks. Existing jailbreak methods typically involve appending carefully crafted prefixes or suffixes to malicious prompts in order to bypass the built-in safety mechanisms of these models. In this work, we identify a new vulnerability in which excessive linguistic complexity can disrupt built-in safety mechanisms-without the need for any added prefixes or suffixes-allowing attackers to elicit harmful outputs directly. We refer to this phenomenon as Information Overload. To automatically exploit this vulnerability, we propose InfoFlood, a jailbreak attack that transforms malicious queries into complex, information-overloaded queries capable of bypassing built-in safety mechanisms. Specifically, InfoFlood: (1) uses linguistic transformations to rephrase malicious queries, (2) identifies the root cause of failure when an attempt is unsuccessful, and (3) refines the prompt's linguistic structure to address the failure while preserving its malicious intent. We empirically validate the effectiveness of InfoFlood on four widely used LLMs-GPT-4o, GPT-3.5-turbo, Gemini 2.0, and LLaMA 3.1-by measuring their jailbreak success rates. InfoFlood consistently outperforms baseline attacks, achieving up to 3 times higher success rates across multiple jailbreak benchmarks. Furthermore, we demonstrate that commonly adopted post-processing defenses, including OpenAI's Moderation API, Perspective API, and SmoothLLM, fail to mitigate these attacks. This highlights a critical weakness in traditional AI safety guardrails when confronted with information overload-based jailbreaks.