LLM-as-a-Judge (LLMaaJ) enables scalable evaluation, yet we lack a decisive test of a judge's qualification: can it recover the hidden objective of a conversation and know when that inference is reliable? Large language models degrade with irrelevant or lengthy context, and multi-turn jailbreaks can scatter goals across turns. We present ObjexMT, a benchmark for objective extraction and metacognition. Given a multi-turn transcript, a model must output a one-sentence base objective and a self-reported confidence. Accuracy is scored by semantic similarity to gold objectives, then thresholded once on 300 calibration items ($τ^\star = 0.66$; $F_1@τ^\star = 0.891$). Metacognition is assessed with expected calibration error, Brier score, Wrong@High-Confidence (0.80 / 0.90 / 0.95), and risk--coverage curves. Across six models (gpt-4.1, claude-sonnet-4, Qwen3-235B-A22B-FP8, kimi-k2, deepseek-v3.1, gemini-2.5-flash) evaluated on SafeMTData\_Attack600, SafeMTData\_1K, and MHJ, kimi-k2 achieves the highest objective-extraction accuracy (0.612; 95\% CI [0.594, 0.630]), while claude-sonnet-4 (0.603) and deepseek-v3.1 (0.599) are statistically tied. claude-sonnet-4 offers the best selective risk and calibration (AURC 0.242; ECE 0.206; Brier 0.254). Performance varies sharply across datasets (16--82\% accuracy), showing that automated obfuscation imposes challenges beyond model choice. High-confidence errors remain: Wrong@0.90 ranges from 14.9\% (claude-sonnet-4) to 47.7\% (Qwen3-235B-A22B-FP8). ObjexMT therefore supplies an actionable test for LLM judges: when objectives are implicit, judges often misinfer them; exposing objectives or gating decisions by confidence is advisable. All experimental data are in the Supplementary Material and at https://github.com/hyunjun1121/ObjexMT_dataset.

While large reasoning models have shown remarkable ability to generate long chains-of-thought (CoTs) in English, we still lack understanding of how these long-form reasoning abilities transfer to the vast majority of the world's languages. In this work, we systematically investigate four key stages of model development--scaling, pretraining, post-training, and inference--to understand how long CoT capabilities extend beyond English. We compare two reasoning settings across nine non-English target languages: En-CoT, where models process target-language inputs, but reason in English; and Target-CoT, where models both process inputs and generate long CoTs in the target language. We find that scaling reasoning model size improves multilingual task performance in En-CoT, but Target-CoT performance lags behind. This gap widens for tasks requiring long, multi-step CoTs such as mathematical reasoning. Shifting to pretraining, we find that adding a specialized reasoning stage enhances En-CoT performance but degrades Target-CoT, whereas broad multilingual pretraining improves both modes simultaneously. Given the scarcity of high-quality reasoning traces in languages other than English, we explore synthetic data curation approaches for post-training. We demonstrate that fine-tuning on reasoning traces automatically translated from gold English traces outperforms fine-tuning on target-language traces distilled from large reasoning models. Finally, we report disparities in inference efficiency between languages and uncover language-specific failure modes in CoTs. We release models, datasets, and code to foster further research.

As program workloads (e.g., AI) increase in size and algorithmic complexity, the primary challenge lies in their high dimensionality, encompassing computing cores, array sizes, and memory hierarchies. To overcome these obstacles, innovative approaches are required. Agile chip design has already benefited from machine learning integration at various stages, including logic synthesis, placement, and routing. With Large Language Models (LLMs) recently demonstrating impressive proficiency in Hardware Description Language (HDL) generation, it is promising to extend their abilities to 2.5D integration, an advanced technique that saves area overhead and development costs. However, LLM-driven chiplet design faces challenges such as flatten design, high validation cost and imprecise parameter optimization, which limit its chiplet design capability. To address this, we propose MAHL, a hierarchical LLM-based chiplet design generation framework that features six agents which collaboratively enable AI algorithm-hardware mapping, including hierarchical description generation, retrieval-augmented code generation, diverseflow-based validation, and multi-granularity design space exploration. These components together enhance the efficient generation of chiplet design with optimized Power, Performance and Area (PPA). Experiments show that MAHL not only significantly improves the generation accuracy of simple RTL design, but also increases the generation accuracy of real-world chiplet design, evaluated by Pass@5, from 0 to 0.72 compared to conventional LLMs under the best-case scenario. Compared to state-of-the-art CLARIE (expert-based), MAHL achieves comparable or even superior PPA results under certain optimization objectives.

Large Language Models (LLMs) are increasingly integrated into real-world applications via the Model Context Protocol (MCP), a universal, open standard for connecting AI agents with data sources and external tools. While MCP enhances the capabilities of LLM-based agents, it also introduces new security risks and expands their attack surfaces. In this paper, we present the first systematic taxonomy of MCP security, identifying 17 attack types across 4 primary attack surfaces. Our benchmark is modular and extensible, allowing researchers to incorporate custom implementations of clients, servers, and transport protocols for systematic security assessment. Experimental results show that over 85% of the identified attacks successfully compromise at least one platform, with core vulnerabilities universally affecting Claude, OpenAI, and Cursor, while prompt-based and tool-centric attacks exhibit considerable variability across different hosts and models. In addition, current protection mechanisms have little effect against these attacks. Large language models (LLMs) are transforming the landscape of intelligent systems, enabling powerful language understanding, reasoning, and generative capabilities. To further unlock their potential in real-world applications, there is an increasing demand for LLMs to interact with external data, tools, and services (Lin et al., 2025; Hasan et al., 2025). The Model Context Protocol (MCP) has emerged as a universal, open standard for connecting AI agents to diverse resources, facilitating richer and more dynamic task-solving. However, this integration also introduces a broader attack surface: vulnerabilities may arise not only from user prompts (such as prompt injection (Shi et al., 2024)), but also from insecure clients, transport protocols, and malicious or misconfigured servers (Hasan et al., 2025). As MCP-powered agents increasingly interact with sensitive enterprise systems and even physical infrastructure, securing the entire MCP stack becomes critical to prevent data breaches, unauthorized actions, and real-world harm (Narajala & Habler, 2025).

Recent years have seen a growing interest and adoption of LLMs, with Mixture-of-Experts (MoE) emerging as a leading architecture in extremely large models. Currently, the largest open-source models reach over $1$T parameters. At such scales, hyperparameter tuning becomes prohibitively expensive. Precisely for this reason, the $μ$Transfer is becoming a key technique. It allows for seamless transfer of optimal hyperparameters across model scales, resulting in a huge reduction in tuning costs. However, existing work has primarily focused on dense LLMs, leaving MoE architectures unexplored. In this work, we derive a $μ$-Parameterization for MoE, providing theoretical guarantees for feature learning across model widths. Our experiments demonstrate that the optimal learning rate reliably transfers across model sizes, establishing a foundation for efficient hyperparameter tuning in large-scale MoE models.

Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs), especially for knowledge-intensive tasks. Despite its advantages, current RAG methods often struggle to fully exploit knowledge during generation. In particular, the synergy between the model's internal parametric knowledge and external retrieved knowledge remains limited. Retrieved contents may sometimes mislead generation, while certain generated content can guide the model toward more accurate outputs. In this work, we propose Collaborative Chain-of-Agents, a framework designed to enhance explicitly synergy over both parametric and retrieved knowledge. Specifically, we first introduce CoCoA-zero, a multi-agent RAG framework that first performs conditional knowledge induction and then reasons answers. Building on this, we develop CoCoA, a long-chain training strategy that synthesizes extended multi-agent reasoning trajectories from CoCoA-zero to fine-tune the LLM. This strategy enhances the model's capability to explicitly integrate and jointly leverage parametric and retrieved knowledge. Experimental results demonstrate the superiority of CoCoA in open-domain QA and multi-hop QA.

Knowledge conflict often arises in retrieval-augmented generation (RAG) systems, where retrieved documents may be inconsistent with one another or contradict the model's parametric knowledge. Existing benchmarks for investigating the phenomenon have notable limitations, including a narrow focus on the question answering setup, heavy reliance on entity substitution techniques, and a restricted range of conflict types. To address these issues, we propose a knowledge graph (KG)-based framework that generates varied and subtle conflicts between two similar yet distinct contexts, while ensuring interpretability through the explicit relational structure of KGs. Experimental results on our benchmark, MAGIC, provide intriguing insights into the inner workings of LLMs regarding knowledge conflict: both open-source and proprietary models struggle with conflict detection -- especially when multi-hop reasoning is required -- and often fail to pinpoint the exact source of contradictions. Finally, we present in-depth analyses that serve as a foundation for improving LLMs in integrating diverse, sometimes even conflicting, information.

Recent studies have shown that Large Language Models (LLMs) are vulnerable to data poisoning attacks, where malicious training examples embed hidden behaviours triggered by specific input patterns. However, most existing works assume a phrase and focus on the attack's effectiveness, offering limited understanding of trigger mechanisms and how multiple triggers interact within the model. In this paper, we present a framework for studying poisoning in LLMs. We show that multiple distinct backdoor triggers can coexist within a single model without interfering with each other, enabling adversaries to embed several triggers concurrently. Using multiple triggers with high embedding similarity, we demonstrate that poisoned triggers can achieve robust activation even when tokens are substituted or separated by long token spans. Our findings expose a broader and more persistent vulnerability surface in LLMs. To mitigate this threat, we propose a post hoc recovery method that selectively retrains specific model components based on a layer-wise weight difference analysis. Our method effectively removes the trigger behaviour with minimal parameter updates, presenting a practical and efficient defence against multi-trigger poisoning.

Large Language Models (LLMs) hold significant promise for transforming digital health by enabling automated medical question answering. However, ensuring these models meet critical industry standards for factual accuracy, usefulness, and safety remains a challenge, especially for open-source solutions. We present a rigorous benchmarking framework using a dataset of over 1,000 health questions. We assess model performance across honesty, helpfulness, and harmlessness. Our results highlight trade-offs between factual reliability and safety among evaluated models -- Mistral-7B, BioMistral-7B-DARE, and AlpaCare-13B. AlpaCare-13B achieves the highest accuracy (91.7%) and harmlessness (0.92), while domain-specific tuning in BioMistral-7B-DARE boosts safety (0.90) despite its smaller scale. Few-shot prompting improves accuracy from 78% to 85%, and all models show reduced helpfulness on complex queries, highlighting ongoing challenges in clinical QA.

Downstream scaling laws aim to predict task performance at larger scales from the model's performance at smaller scales. Whether such prediction should be possible is unclear: some works discover clear linear scaling trends after simple transformations of the performance metric, whereas others point out fundamental challenges to downstream scaling laws, such as emergence and inverse scaling. In this work, we conduct a meta-analysis of existing data on downstream scaling laws, and we find that predictable scaling only occurs in a minority of cases: 39% of the time. Moreover, seemingly benign changes to the experimental setting can completely change the scaling behavior. Our analysis underscores the need to understand the conditions under which scaling laws succeed. To accurately model the relationship between pretraining loss and task performance, we must embrace the cases in which scaling behavior deviates from linear trends.