--With the rapid advent of generative models, efficiently deploying these models on specialized hardware has become critical. T ensor Processing Units (TPUs) are designed to accelerate AI workloads, but their high power consumption necessitates innovations for improving efficiency. Compute-in-memory (CIM) has emerged as a promising paradigm with superior area and energy efficiency. In this work, we present a TPU architecture that integrates digital CIM to replace conventional digital systolic arrays in matrix multiply units (MXUs). We first establish a CIM-based TPU architecture model and simulator to evaluate the benefits of CIM for diverse generative model inference. Building upon the observed design insights, we further explore various CIM-based TPU architectural design choices. Up to 44.2% and 33.8% performance improvement for large language model and diffusion transformer inference, and 27.3 reduction in MXU energy consumption can be achieved with different design choices, compared to the baseline TPUv4i architecture. Generative models, such as large language models (LLMs) and diffusion models (DMs), have exhibited exceptional performance in generating content across various modalities. For example, LLMs have dominated NLP tasks, powering applications like ChatGPT [1].

Existing Existing automatic audio generation methods struggle to generate podcast-like audio programs effectively. The key challenges lie in in-depth content generation, appropriate and expressive voice production. This paper proposed PodAgent, a comprehensive framework for creating audio programs. PodAgent 1) generates informative topic-discussion content by designing a Host-Guest-Writer multi-agent collaboration system, 2) builds a voice pool for suitable voice-role matching and 3) utilizes LLM-enhanced speech synthesis method to generate expressive conversational speech. Given the absence of standardized evaluation criteria for podcast-like audio generation, we developed comprehensive assessment guidelines to effectively evaluate the model's performance. Experimental results demonstrate PodAgent's effectiveness, significantly surpassing direct GPT-4 generation in topic-discussion dialogue content, achieving an 87.4% voice-matching accuracy, and producing more expressive speech through LLM-guided synthesis. Demo page: https://podcast-agent.github.io/demo/. Source code: https://github.com/yujxx/PodAgent.

Research on metaphor has steadily increased over the last decades, as this phenomenon opens a window into a range of processes in language and cognition, from pragmatic inference to abstraction and embodied simulation. At the same time, the demand for rigorously constructed and extensively normed experimental materials increased as well. Here, we present the Figurative Archive, an open database of 997 metaphors in Italian enriched with rating and corpus-based measures (from familiarity to lexical frequency), derived by collecting stimuli used across 11 studies. It includes both everyday and literary metaphors, varying in structure and semantic domains. Dataset validation comprised correlations between familiarity and other measures. The Figurative Archive has several aspects of novelty: it is increased in size compared to previous resources; it includes a novel measure of inclusiveness, to comply with current recommendations for non-discriminatory language use; it is displayed in a web-based interface, with features for a flexible and customized consultation. We provide guidelines for using the Archive in future metaphor studies, in the spirit of open science.

In the dynamic landscape of artificial intelligence, the exploration of hallucinations within vision-language (VL) models emerges as a critical frontier. This work delves into the intricacies of hallucinatory phenomena exhibited by widely used image captioners, unraveling interesting patterns. Specifically, we step upon previously introduced techniques of conceptual counterfactual explanations to address VL hallucinations. The deterministic and efficient nature of the employed conceptual counterfactuals backbone is able to suggest semantically minimal edits driven by hierarchical knowledge, so that the transition from a hallucinated caption to a non-hallucinated one is performed in a black-box manner. HalCECE, our proposed hallucination detection framework is highly interpretable, by providing semantically meaningful edits apart from standalone numbers, while the hierarchical decomposition of hallucinated concepts leads to a thorough hallucination analysis. Another novelty tied to the current work is the investigation of role hallucinations, being one of the first works to involve interconnections between visual concepts in hallucination detection. Overall, HalCECE recommends an explainable direction to the crucial field of VL hallucination detection, thus fostering trustworthy evaluation of current and future VL systems.

This paper aims to present new alarming trends in the field of child sexual abuse through imagery, as part of SafeLine's research activities in the field of cybercrime, child sexual abuse material and the protection of children's rights to safe online experiences. It focuses primarily on the phenomenon of AI-generated CSAM, sophisticated ways employed for its production which are discussed in dark web forums and the crucial role that the open-source AI models play in the evolution of this overwhelming phenomenon. The paper's main contribution is a correlation analysis between the hotline's reports and domain names identified in dark web forums, where users' discussions focus on exchanging information specifically related to the generation of AI-CSAM. The objective was to reveal the close connection of clear net and dark web content, which was accomplished through the use of the ATLAS dataset of the Voyager system. Furthermore, through the analysis of a set of posts' content drilled from the above dataset, valuable conclusions on forum members' techniques employed for the production of AI-generated CSAM are also drawn, while users' views on this type of content and routes followed in order to overcome technological barriers set with the aim of preventing malicious purposes are also presented. As the ultimate contribution of this research, an overview of the current legislative developments in all country members of the INHOPE organization and the issues arising in the process of regulating the AI- CSAM is presented, shedding light in the legal challenges regarding the regulation and limitation of the phenomenon.

Polarization is defined as divisive opinions held by two or more groups on substantive issues. As the world's third-largest democracy, Indonesia faces growing concerns about the interplay between political polarization and online toxicity, which is often directed at vulnerable minority groups. Despite the importance of this issue, previous NLP research has not fully explored the relationship between toxicity and polarization. To bridge this gap, we present a novel multi-label Indonesian dataset that incorporates toxicity, polarization, and annotator demographic information. Benchmarking this dataset using BERT-base models and large language models (LLMs) shows that polarization information enhances toxicity classification, and vice versa. Furthermore, providing demographic information significantly improves the performance of polarization classification.

Multimodal large language models (MLLMs) have garnered widespread attention from researchers due to their remarkable understanding and generation capabilities in visual language tasks (e.g., visual question answering). However, the rapid pace of knowledge updates in the real world makes offline training of MLLMs costly, and when faced with non-stationary data streams, MLLMs suffer from catastrophic forgetting during learning. In this paper, we propose an MLLMs-based dual momentum Mixture-of-Experts (CL-MoE) framework for continual visual question answering (VQA). We integrate MLLMs with continual learning to utilize the rich commonsense knowledge in LLMs. We introduce a Dual-Router MoE (RMoE) strategy to select the global and local experts using task-level and instance-level routers, to robustly assign weights to the experts most appropriate for the task. Then, we design a dynamic Momentum MoE (MMoE) to update the parameters of experts dynamically based on the relationships between the experts and tasks/instances, so that the model can absorb new knowledge while maintaining existing knowledge. The extensive experimental results indicate that our method achieves state-of-the-art performance on 10 VQA tasks, proving the effectiveness of our approach.

Processing long contexts has become a critical capability for modern large language models (LLMs). However, serving long-context LLMs comes with significant inference costs due to the high memory overhead of the key-value (KV) cache. Existing work leverages dynamic sparse attention algorithms (DSAes) to mitigate the KV cache overhead, but these algorithms rely on top-$k$ KV cache selection, which results in a trade-off between accuracy and efficiency. A larger $k$ improves accuracy but decreases efficiency, while a smaller $k$ boosts efficiency but compromises accuracy. To overcome this trade-off, this paper presents PSA, a $\underline{P}$rogressive $\underline{S}$parse $\underline{A}$ttention mechanism that integrates algorithmic innovations with system co-design to achieve both high inference accuracy and improved efficiency in LLM serving. The PSA algorithm adaptively adjusts the KV cache budget of different tokens and layers according to their real attention weight distributions, rather than relying on a fixed budget $k$. This enables high accuracy while minimizing KV cache usage. To further enhance execution efficiency, we introduce a pipelined iteration scheme that reduces CPU-GPU interleaving and synchronization overhead during PSA computation. Additionally, we implement unified GPU memory management that optimizes PSA's memory utilization by accounting for uneven memory requirements across different model layers. Extensive experimental results demonstrate that PSA reduces KV cache usage for attention computation by up to 2.4$\times$ and 8.8$\times$, and increases end-to-end serving throughput by up to 1.4$\times$ and 2.0$\times$, compared to state-of-the-art DSAes and systems without sparse attention, respectively.

From disinformation spread by AI chatbots to AI recommendations that inadvertently reinforce stereotypes, textual bias poses a significant challenge to the trustworthiness of large language models (LLMs). In this paper, we propose a multi-agent framework that systematically identifies biases by disentangling each statement as fact or opinion, assigning a bias intensity score, and providing concise, factual justifications. Evaluated on 1,500 samples from the WikiNPOV dataset, the framework achieves 84.9% accuracy$\unicode{x2014}$an improvement of 13.0% over the zero-shot baseline$\unicode{x2014}$demonstrating the efficacy of explicitly modeling fact versus opinion prior to quantifying bias intensity. By combining enhanced detection accuracy with interpretable explanations, this approach sets a foundation for promoting fairness and accountability in modern language models.

Recent advancements in Large Language Models (LLMs) have expanded their context windows to unprecedented lengths, sparking debates about the necessity of Retrieval-Augmented Generation (RAG). To address the fragmented evaluation paradigms and limited cases in existing Needle-in-a-Haystack (NIAH), this paper introduces U-NIAH, a unified framework that systematically compares LLMs and RAG methods in controlled long context settings. Our framework extends beyond traditional NIAH by incorporating multi-needle, long-needle, and needle-in-needle configurations, along with different retrieval settings, while leveraging the synthetic Starlight Academy dataset-a fictional magical universe-to eliminate biases from pre-trained knowledge. Through extensive experiments, we investigate three research questions: (1) performance trade-offs between LLMs and RAG, (2) error patterns in RAG, and (3) RAG's limitations in complex settings. Our findings show that RAG significantly enhances smaller LLMs by mitigating the "lost-in-the-middle" effect and improving robustness, achieving an 82.58% win-rate over LLMs. However, we observe that retrieval noise and reverse chunk ordering degrade performance, while surprisingly, advanced reasoning LLMs exhibit reduced RAG compatibility due to sensitivity to semantic distractors. We identify typical error patterns including omission due to noise, hallucination under high noise critical condition, and self-doubt behaviors. Our work not only highlights the complementary roles of RAG and LLMs, but also provides actionable insights for optimizing deployments. Code: https://github.com/Tongji-KGLLM/U-NIAH.