March 2026

arXiv:2603.09988v1 Announce Type: new Abstract: Mechanistic interpretability identifies internal circuits responsible for model behaviors, yet translating these findings into human-understandable explanations remains an open problem. We present a pipeline that bridges circuit-level analysis and natural language explanations by (i) identifying causally important attention heads via activation patching, (ii) generating explanations using both template-based and LLM-based methods, and (iii) evaluating faithfulness using ERASER-style metrics adapted for circuit-level attribution. We evaluate on the Indirect Object Identification (IOI) task in GPT-2 […]

Ver mais

arXiv:2603.09987v1 Announce Type: new Abstract: Feature Transformation (FT) is a core data-centric AI task that improves feature space quality to advance downstream predictive performance. However, discovering effective transformations remains challenging due to the large space of feature-operator combinations. Existing solutions rely on discrete search or latent generation, but they are frequently limited by sample inefficiency, invalid candidates, and redundant generations with limited coverage. Large Language Models (LLMs) offer strong priors for producing valid transformations, but current LLM-based FT […]

Ver mais

arXiv:2603.09986v1 Announce Type: new Abstract: Hallucinations, the tendency for large language models to provide responses with factually incorrect and unsupported claims, is a serious problem within natural language processing for which we do not yet have an effective solution to mitigate against. Existing benchmarks for medical QA rarely evaluate this behavior against a fixed evidence source. We ask how often hallucinations occur on textbook-grounded QA and how responses to medical QA prompts vary across models. We conduct two […]

Ver mais

arXiv:2603.09985v1 Announce Type: new Abstract: Large language models (LLMs) have demonstrated remarkable capabilities across diverse tasks, yet their ability to accurately assess their own confidence remains poorly understood. We present an empirical study investigating whether LLMs exhibit patterns reminiscent of the Dunning-Kruger effect — a cognitive bias where individuals with limited competence tend to overestimate their abilities. We evaluate four state-of-the-art models (Claude Haiku 4.5, Gemini 2.5 Pro, Gemini 2.5 Flash, and Kimi K2) across four benchmark datasets […]

Ver mais

arXiv:2603.09984v1 Announce Type: new Abstract: The digital age has expanded social media and online forums, allowing free expression for nearly 45% of the global population. Yet, it has also fueled online harassment, bullying, and harmful behaviors like hate speech and toxic comments across social networks, messaging apps, and gaming communities. Studies show 65% of parents notice hostile online behavior, and one-third of adolescents in mobile games experience bullying. A substantial volume of abusive content is generated and shared […]

Ver mais

arXiv:2603.09983v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) models enable scalable performance but face severe memory constraints on edge devices. Existing offloading strategies struggle with I/O bottlenecks due to the dynamic, low-information nature of autoregressive expert activation. In this paper, we propose to repurpose Speculative Decoding (SD) not merely as a compute accelerator, but as an informative lookahead sensor for memory management, supported by our theoretical and empirical analyses. Hence, we introduce MoE-SpAc, an MoE inference framework that integrates […]

Ver mais

arXiv:2603.09981v1 Announce Type: new Abstract: Summarization is a core task in Natural Language Processing (NLP). Recent advances in Large Language Models (LLMs) and the introduction of large context windows reaching millions of tokens make it possible to process entire books in a single prompt. At the same time, for well-known books, LLMs can generate summaries based only on internal knowledge acquired during training. This raises several important questions: How do summaries generated from internal memory compare to those […]

Ver mais

arXiv:2603.09980v1 Announce Type: new Abstract: LLM unlearning is essential for mitigating safety, copyright, and privacy concerns in pre-trained large language models (LLMs). Compared to preference alignment, it offers a more explicit way by removing undesirable knowledge characterized by specific unlearning datasets. In previous works, gradient ascent (GA) and its variants have shown promise for implementing unlearning, yet their untargeted nature results in unintended degradation of general capabilities, incomplete removal of knowledge, and the generation of incoherent responses, among […]

Ver mais

arXiv:2603.09979v1 Announce Type: new Abstract: Persian poetry plays an active role in Iranian cultural practice, where verses by canonical poets such as Hafez are frequently quoted, paraphrased, or completed from partial cues. Supporting such interactions requires language models to engage not only with poetic meaning but also with culturally entrenched surface form. We introduce GhazalBench, a benchmark for evaluating how large language models (LLMs) interact with Persian ghazals under usage-grounded conditions. GhazalBench assesses two complementary abilities: producing faithful […]

Ver mais

arXiv:2603.09978v1 Announce Type: new Abstract: Large language models have recently surpassed specialized systems on code generation, yet their effectiveness on other code-analysis tasks remains less clear. At the same time, multi-task learning offers a way to unify diverse objectives within a single model, but fully fine-tuning LLMs across tasks is computationally prohibitive. Parameter-efficient fine-tuning mitigates this cost by updating only a small fraction of weights. Although PEFT has proven effective in single-task settings, its potential for multi-task learning […]

Ver mais