Large Language Models (LLMs), especially their compact efficiency-oriented variants, remain susceptible to jailbreak attacks that can elicit harmful outputs despite extensive alignment efforts. Existing adversarial prompt generation techniques often rely on manual engineering or rudimentary obfuscation, producing low-quality or incoherent text that is easily flagged by perplexity-based filters. We present an automated red-teaming framework that evolves semantically meaningful and stealthy jailbreak prompts for aligned compact LLMs. The approach employs a multi-stage evolutionary search, where candidate prompts are iteratively refined using a population-based strategy augmented with temperature-controlled variability to balance exploration and coherence preservation. This enables the systematic discovery of prompts capable of bypassing alignment safeguards while maintaining natural language fluency. We evaluate our method on benchmarks in English (In-The-Wild Jailbreak Prompts on LLMs), and a newly curated Arabic one derived from In-The-Wild Jailbreak Prompts on LLMs and annotated by native Arabic linguists, enabling multilingual assessment.

The advent of Large Language Models (LLMs) has provided unprecedented capabilities for analyzing unstructured text data. However, deploying these models as reliable, robust, and scalable classifiers in production environments presents significant methodological challenges. Standard fine-tuning approaches can be resource-intensive and often struggle with the dynamic nature of real-world data distributions, which is common in the industry. In this paper, we propose a comprehensive, semi-supervised framework that leverages the zero- and few-shot capabilities of LLMs for building hierarchical text classifiers as a framework for a solution to these industry-wide challenges. Our methodology emphasizes an iterative, human-in-the-loop process that begins with domain knowledge elicitation and progresses through prompt refinement, hierarchical expansion, and multi-faceted validation. We introduce techniques for assessing and mitigating sequence-based biases and outline a protocol for continuous monitoring and adaptation. This framework is designed to bridge the gap between the raw power of LLMs and the practical need for accurate, interpretable, and maintainable classification systems in industry applications.

Large Language Models (LLMs) have demonstrated the capability to refine their generated answers through self-correction, enabling continuous performance improvement over multiple rounds. However, the mechanisms underlying how and why accuracy evolves during this iterative process remain unexplored. To fill this gap, we propose a probabilistic theory to model the dynamics of accuracy change and explain the performance improvements observed in multi-round self-correction. Through mathematical derivation, we establish that the accuracy after the $t^{th}$ round of self-correction is given by: $Acc_t = Upp - α^t(Upp - Acc_0),$ where $Acc_0$ denotes the initial accuracy, $Upp$ represents the upper bound of accuracy convergence, and $α$ determines the rate of convergence. Based on our theory, these parameters can be calculated and the predicted accuracy curve then can be obtained through only a single round of self-correction. Extensive experiments across diverse models and datasets demonstrate that our theoretical predictions align closely with empirical accuracy curves, validating the effectiveness of the theory. Our work provides a theoretical foundation for understanding LLM self-correction, thus paving the way for further explorations.

Over the past decade, adaptive video streaming technology has witnessed significant advancements, particularly driven by the rapid evolution of deep learning techniques. However, the black-box nature of deep learning algorithms presents challenges for developers in understanding decision-making processes and optimizing for specific application scenarios. Although existing research has enhanced algorithm interpretability through decision tree conversion, interpretability does not directly equate to developers' subjective comprehensibility. To address this challenge, we introduce \texttt{ComTree}, the first bitrate adaptation algorithm generation framework that considers comprehensibility. The framework initially generates the complete set of decision trees that meet performance requirements, then leverages large language models to evaluate these trees for developer comprehensibility, ultimately selecting solutions that best facilitate human understanding and enhancement. Experimental results demonstrate that \texttt{ComTree} significantly improves comprehensibility while maintaining competitive performance, showing potential for further advancement. The source code is available at https://github.com/thu-media/ComTree.

We introduce Cetvel, a comprehensive benchmark designed to evaluate large language models (LLMs) in Turkish. Existing Turkish benchmarks often lack either task diversity or culturally relevant content, or both. Cetvel addresses these gaps by combining a broad range of both discriminative and generative tasks ensuring content that reflects the linguistic and cultural richness of Turkish language. Cetvel covers 23 tasks grouped into seven categories, including tasks such as grammatical error correction, machine translation, and question answering rooted in Turkish history and idiomatic language. We evaluate 33 open-weight LLMs (up to 70B parameters) covering different model families and instruction paradigms. Our experiments reveal that Turkish-centric instruction-tuned models generally underperform relative to multilingual or general-purpose models (e.g. Llama 3 and Mistral), despite being tailored for the language. Moreover, we show that tasks such as grammatical error correction and extractive question answering are particularly discriminative in differentiating model capabilities. Cetvel offers a comprehensive and culturally grounded evaluation suite for advancing the development and assessment of LLMs in Turkish.

Large Language Models (LLMs) such as ChatGPT-4, Claude 3, and LLaMA 4 are increasingly embedded in software/application development, supporting tasks from code generation to debugging. Yet, their real-world effectiveness in detecting diverse software bugs, particularly complex, security-relevant vulnerabilities, remains underexplored. This study presents a systematic, empirical evaluation of these three leading LLMs using a benchmark of foundational programming errors, classic security flaws, and advanced, production-grade bugs in C++ and Python. The dataset integrates real code from SEED Labs, OpenSSL (via the Suresoft GLaDOS database), and PyBugHive, validated through local compilation and testing pipelines. A novel multi-stage, context-aware prompting protocol simulates realistic debugging scenarios, while a graded rubric measures detection accuracy, reasoning depth, and remediation quality. Our results show that all models excel at identifying syntactic and semantic issues in well-scoped code, making them promising for educational use and as first-pass reviewers in automated code auditing. Performance diminishes in scenarios involving complex security vulnerabilities and large-scale production code, with ChatGPT-4 and Claude 3 generally providing more nuanced contextual analyses than LLaMA 4. This highlights both the promise and the present constraints of LLMs in serving as reliable code analysis tools.

Competitive programming has emerged as a critical benchmark for evaluating the reasoning and coding capabilities of Large Language Models (LLMs). Despite impressive progress on existing benchmarks, we argue that current evaluations overstate model proficiency, masking a substantial gap between LLMs and elite human programmers. This gap arises from two key limitations: insufficient difficulty and scope of benchmark problems, and evaluation bias from low-quality test cases. To address these shortcomings, we present AetherCode, a new benchmark that draws problems from premier programming competitions such as IOI and ICPC, offering broader coverage and higher difficulty. AetherCode further incorporates comprehensive, expert-validated test suites built through a hybrid of automated generation and human curation, ensuring rigorous and reliable assessment. By combining challenging problem design with robust evaluation, AetherCode provides a more faithful measure of LLM capabilities and sets a new standard for future research in code reasoning.

Audio-driven facial animation presents an effective solution for animating digital avatars. In this paper, we detail the technical aspects of NVIDIA Audio2Face-3D, including data acquisition, network architecture, retargeting methodology, evaluation metrics, and use cases. Audio2Face-3D system enables real-time interaction between human users and interactive avatars, facilitating facial animation authoring for game characters. To assist digital avatar creators and game developers in generating realistic facial animations, we have open-sourced Audio2Face-3D networks, SDK, training framework, and example dataset.

Large Language Models can emulate different writing styles, ranging from composing poetry that appears indistinguishable from that of famous poets to using slan g that can convince people that they are chatting with a human online . While differences in style may not always be visible to the untrained eye, we can generally distinguish the writing of different people, like a linguistic fingerprint. This work examines whether a language model can also be linked to a specific fingerprint . Through stylometric and multidimensional register analys e s, w e compare human - authored and model - authored texts from different registers. We find that the model can successfully adapt its style depending on whether it is prompted to produce a Wikipedia entry vs. a college essay, but not in a way that makes it indistinguishable from human s . Concretely, the model shows more limited variation when producing outputs in different registers. O ur results suggest that the model prefers nouns to verbs, thus showing a distinct linguistic backbone from humans, who tend to anchor language in the highly grammaticalized dimensions of tense, aspect, and mood . It is possible that the more complex domains of grammar reflect a mode of thought unique to humans, thus acting as a litmus test for Artificial Intelligence. 2 Introduction Scholars from different disciplines have been addressing the question of what makes us human for centuries. For Nobel laureate Bertrand Russell, the answer is language, for "no matter how eloquently a dog may bark, he cannot tell you that his parents were poor but honest". H uman language is both flexible and constrained at the same time, and this is why the Turing Test, described as a litmus test for Artificial Intelligence [ Shieber 199 4, French 200 0], is linked to achieving a level of conversational proficiency that is highly complex, akin to that of a human [ Turing 1950 ] . Human language is flexible in the sense that we all make different choices when conversing. Every human is thought t o have a distinct linguistic fingerprint called idiolect [ Halliday et al. 196 4, Coulthard 2004 ] . This idiolect, which can be defined as an individual's unique use of linguistic forms (including lexical choices, collocations and fixed expressions, punctuation patterns, misspellings, and grammatical style), is critical for authorship attribution in a range of situations: from identifying that a poem with dashes, elliptical syntax, and unconventional capitalization is more likely authored by Emily Dickinson and not by William Shakespeare, to pinning down a person of interest in the course of a criminal investigation, as happened in the Unabomber case .

Legal judgment prediction (LJP) has become increasingly important in the legal field. In this paper, we identify that existing large language models (LLMs) have significant problems of insufficient reasoning due to a lack of legal knowledge. Therefore, we introduce GLARE, an agentic legal reasoning framework that dynamically acquires key legal knowledge by invoking different modules, thereby improving the breadth and depth of reasoning. Experiments conducted on the real-world dataset verify the effectiveness of our method. Furthermore, the reasoning chain generated during the analysis process can increase interpretability and provide the possibility for practical applications.