Tootfinder

Multi-Physics: A Comprehensive Benchmark for Multimodal LLMs Reasoning on Chinese Multi-Subject Physics Problems
Zhongze Luo, Zhenshuai Yin, Yongxin Guo, Zhichao Wang, Jionghao Zhu, Xiaoying Tang
https://arxiv.org/abs/2509.15839

Multi-Physics: A Comprehensive Benchmark for Multimodal LLMs Reasoning on Chinese Multi-Subject Physics Problems
While multimodal LLMs (MLLMs) demonstrate remarkable reasoning progress, their application in specialized scientific domains like physics reveals significant gaps in current evaluation benchmarks. Specifically, existing benchmarks often lack fine-grained subject coverage, neglect the step-by-step reasoning process, and are predominantly English-centric, failing to systematically evaluate the role of visual information. Therefore, we introduce \textbf {Multi-Physics} for Chinese physics reasonin…

Jordanian spin chains for twisted strings in $AdS_5\times S^5$
Sibylle Driezen, Adrien Molines
https://arxiv.org/abs/2507.13911 https://

Jordanian spin chains for twisted strings in $AdS_5\times S^5$
We study the integrable spin chain formulation of Jordanian deformations of the $AdS_5\times S^5$ superstring, realised via Drinfel'd twists. Among these models, we first identify a unique supergravity deformation confined to an $SL(2,\mathbb{R})$ sector and with constant dilaton. We then develop a general framework for closed Drinfel'd twisted spin chains and construct an explicit map to undeformed models with twisted-boundary conditions. Applied to the non compact $\mathrm{XXX}_{-1/2}$ spin c…

Browns not mulling QB change despite struggles https://www.espn.com/nfl/story/_/id/46261767/coach-browns-not-mulling-qb-change-joe-flacco-struggles

HumanPCR: Probing MLLM Capabilities in Diverse Human-Centric Scenes
Keliang Li, Hongze Shen, Hao Shi, Ruibing Hou, Hong Chang, Jie Huang, Chenghao Jia, Wen Wang, Yiling Wu, Dongmei Jiang, Shiguang Shan, Xilin Chen
https://arxiv.org/abs/2508.13692

HumanPCR: Probing MLLM Capabilities in Diverse Human-Centric Scenes
The aspiration for artificial general intelligence, fueled by the rapid progress of multimodal models, demands human-comparable performance across diverse environments. We propose HumanPCR, an evaluation suite for probing MLLMs' capacity about human-related visual contexts across three hierarchical levels: Perception, Comprehension, and Reasoning (denoted by Human-P, Human-C, and Human-R, respectively). Human-P and Human-C feature over 6,000 human-verified multiple choice questions, assessing m…

Gorgeous: Revisiting the Data Layout for Disk-Resident High-Dimensional Vector Search
Peiqi Yin, Xiao Yan, Qihui Zhou, Hui Li, Xiaolu Li, Lin Zhang, Meiling Wang, Xin Yao, James Cheng
https://arxiv.org/abs/2508.15290

Gorgeous: Revisiting the Data Layout for Disk-Resident High-Dimensional Vector Search
Similarity-based vector search underpins many important applications, but a key challenge is processing massive vector datasets (e.g., in TBs). To reduce costs, some systems utilize SSDs as the primary data storage. They employ a proximity graph, which connects similar vectors to form a graph and is the state-of-the-art index for vector search. However, these systems are hindered by sub-optimal data layouts that fail to effectively utilize valuable memory space to reduce disk access and suffer …

Unveiling Trust in Multimodal Large Language Models: Evaluation, Analysis, and Mitigation
Yichi Zhang, Yao Huang, Yifan Wang, Yitong Sun, Chang Liu, Zhe Zhao, Zhengwei Fang, Huanran Chen, Xiao Yang, Xingxing Wei, Hang Su, Yinpeng Dong, Jun Zhu
https://arxiv.org/abs/2508.15370

Unveiling Trust in Multimodal Large Language Models: Evaluation, Analysis, and Mitigation
The trustworthiness of Multimodal Large Language Models (MLLMs) remains an intense concern despite the significant progress in their capabilities. Existing evaluation and mitigation approaches often focus on narrow aspects and overlook risks introduced by the multimodality. To tackle these challenges, we propose MultiTrust-X, a comprehensive benchmark for evaluating, analyzing, and mitigating the trustworthiness issues of MLLMs. We define a three-dimensional framework, encompassing five trustwo…

"""
In melancholy, the spirits are carried away by an agitation, but a weak agitation that lacks power or violence, a sort of impotent upset that follows neither a particular path nor the aperta opercula [open ways], but traverses the cerebral matter constantly creating new pores. Yet the spirits do not wander far on the new paths they create, and their agitation dies down rapidly, as their strength is quickly spent and motion comes to a halt: ‘non longe perveniunt’ [they do not reach far]. A trouble of this nature, common to all delirium, does not have the power to produce on the surface of the body the violent movements or the cries to be observed in mania and frenzy. Melancholy never attains frenzy; it is a madness always at the limits of its own impotence. That paradox is explained by the secret alterations in the spirits. Ordinarily, they travel with the speed and instantaneous transparency of rays of light, but in melancholy they become weighed down with night, becoming ‘obscure, thick and dark’, and the images of things that they bring before consciousness are ‘in a shadow, or covered with darkness’. As a result they move more slowly, and are more like a dark, chemical vapour than pure light. This chemical vapour is acid in nature, rather than sulphurous or alcoholic, for in acid vapours the particles are mobile and incapable of repose, but their activity is weak and without consequence. When they are distilled, all that remains in the still is a kind of insipid phlegm. Acid vapours, therefore, are taken to have the same properties as melancholy, whereas alcoholic vapours, which are always ready to burst into flames, are more related to frenzy, and sulphurous vapours bring on mania, as they are agitated by continuous, violent movement. If the ‘formal reason and causes’ of melancholy were to be sought, it made sense to look for them in the vapours that rose up from the blood to the head, and which had degenerated into ‘an acetous or sharp distillation’. A cursory glance seems to indicate that a melancholy of spirits and a whole chemistry of humours lies behind Willis’ analyses, but in fact his guiding principle mostly reflects the immediate qualities of the melancholic illness: an impotent disorder, and the shadow that comes over the spirit with an acrid acidity that slowly corrodes the heart and the mind. The chemistry of acids is not an explanation of the symptoms, but a qualitative option: a whole phenomenology of melancholic experience.
"""
(Michel Foucault, History of Madness)

When Safe Unimodal Inputs Collide: Optimizing Reasoning Chains for Cross-Modal Safety in Multimodal Large Language Models
Wei Cai, Shujuan Liu, Jian Zhao, Ziyan Shi, Yusheng Zhao, Yuchen Yuan, Tianle Zhang, Chi Zhang, Xuelong Li
https://arxiv.org/abs/2509.12060

When Safe Unimodal Inputs Collide: Optimizing Reasoning Chains for Cross-Modal Safety in Multimodal Large Language Models
Multimodal Large Language Models (MLLMs) are susceptible to the implicit reasoning risk, wherein innocuous unimodal inputs synergistically assemble into risky multimodal data that produce harmful outputs. We attribute this vulnerability to the difficulty of MLLMs maintaining safety alignment through long-chain reasoning. To address this issue, we introduce Safe-Semantics-but-Unsafe-Interpretation (SSUI), the first dataset featuring interpretable reasoning paths tailored for such a cross-modal c…

YouTube is like heroin because it pulls you in, hooks you, and slowly drains you. It is designed to be addictive and destructive, and that is exactly why it feels so hard to step away from it.
And yes, I know the excuse, maybe your favorite dealer has not set up shop on PeerTube. But that does not change the truth of the comparison.
PeerTube is like weed, it grows more naturally, it is gentler on you, and it still gives you a high without tearing apart your health and your lif…

Don't Change My View: Ideological Bias Auditing in Large Language Models
Paul Kr\"oger, Emilio Barkett
https://arxiv.org/abs/2509.12652 https://ar…

Don't Change My View: Ideological Bias Auditing in Large Language Models
As large language models (LLMs) become increasingly embedded in products used by millions, their outputs may influence individual beliefs and, cumulatively, shape public opinion. If the behavior of LLMs can be intentionally steered toward specific ideological positions, such as political or religious views, then those who control these systems could gain disproportionate influence over public discourse. Although it remains an open question whether LLMs can reliably be guided toward coherent ide…