Tootfinder

KIS-S: A GPU-Aware Kubernetes Inference Simulator with RL-Based Auto-Scaling
Guilin Zhang, Wulan Guo, Ziqi Tan, Qiang Guan, Hailong Jiang
https://arxiv.org/abs/2507.07932

KIS-S: A GPU-Aware Kubernetes Inference Simulator with RL-Based Auto-Scaling
Autoscaling GPU inference workloads in Kubernetes remains challenging due to the reactive and threshold-based nature of default mechanisms such as the Horizontal Pod Autoscaler (HPA), which struggle under dynamic and bursty traffic patterns and lack integration with GPU-level metrics. We present KIS-S, a unified framework that combines KISim, a GPU-aware Kubernetes Inference Simulator, with KIScaler, a Proximal Policy Optimization (PPO)-based autoscaler. KIScaler learns latency-aware and resour…

Scaling laws for doublet craters formed by low-velocity impacts of unequal-mass spheres into a granular bed
Haruto Kitagawa, Ririha Obara, Yu Matsuda
https://arxiv.org/abs/2507.04593

Scaling laws for doublet craters formed by low-velocity impacts of unequal-mass spheres into a granular bed
Understanding the formation mechanism of doublet craters is an important challenge for advancing knowledge in astronomy and granular physics. In this study, we investigated craters formed by low-velocity impacts of two steel spheres with different masses into a granular bed. Even when the masses were different, a figure-eight-shaped doublet crater and a central ridge were observed, similar to the case with equal masses. However, the resulting shape became asymmetric even without a time delay be…

Sometimes I daydream of an Anarchist navy, silently sailing across the Salish Sea, smuggling to avoid the tariffs and annoy the terfs. I dream of them sneaking by CPB, who is then deployed with the rest of the federal forces to occupy cities, in stolen ships repaired with materials stolen from prison construction sites... and with the sale of tariff skipped good, they fund a Cascadian version of the AIF.
And this dream brings a smile to my face.
Good luck to anyone who is inspired to join the Anarchist Yacht Club.

Ukraine verlässt Abkommen über Minen-Verbot
Nach den baltischen Staaten will auch die Ukraine aus der Ottawa-Konvention zum Verbot von Antipersonenminen austreten. Präsident Wolodymyr Selenskyj habe ein entsprechendes Dekret unterzeichnet, teilte der ukrainische Parlamentsabgeordnete Roman Kostenko bei Facebook mit. Das Dokument wurde am Nachmittag auch vom Präsidentenamt in Kiew veröffentlicht. Zuv…
🔗

Landminen in der Ukraine: Auf Jahrzehnte vermint
In der Ukraine soll eine Fläche von der doppelten Größe Österreichs mit Landminen verseucht sein - wohl auf Jahrzehnte. In manchen Orten kann jeder Schritt furchtbare Verletzungen bedeuten.

“Are we the baddies?” is not a question anyone in the West has to ask themselves. There’s no question about it. You were. You are. Will you continue to be? That’s the only question that matters and the answer is entirely up to you.
Silence is complicity.
Inaction is complicity.
Apologism is complicity.
Do something!
#israel

End Israeli Apartheid (@EndIsraeliApartheid@mastodon.social)
Attached: 1 image More RAF jets are being sent to the Middle East, Keir Starmer has said after Iran threatened to target UK, French and US bases if the countries help stop strikes on Israel. Additional refuelling aircraft have been deployed from UK bases and more fast jets will be sent over, it is understood. The UK already has RAF jets in the Middle East as part of Operation Shader. https://www.thenational.scot/news/25239897.uk-jets-sent-middle-east-despite-calls-de-escalation/ 🕎 �…

Imaginary scaling
P. M. Ferreira, B. Grzadkowski, O. M. Ogreid
https://arxiv.org/abs/2506.21145 https://arxiv.org/pdf/2506.21145

Imaginary scaling
Recently, a hitherto unknown class of RG-stable relations between parameters of bosonic field theories have been identified and dubbed as the r0 or 'GOOFy' symmetries. Here, 1-loop properties of the r0 invariant 2-Higgs Doublet Model and a minimal symmetric model are discussed. It is concluded that the symmetry is present at the 1-loop provided the UV cut-off squared transforms non-trivially under r0. The minimal model requires the presence of two real fields.

Happy SysAdmin Day everyone! Here is a head scratcher. Website kept serving the default new site page. Even when I made a new index.html file. So I made a test.html and got Forbidden. Error log says "client denied by server configuration: /var/www/newwebsite/test.html" Not directory permissions and not selinux. So I made /var/www/newweb2 and moved the files over. Voila, working. Both test.htnl and new index.html. I'm leaving it alone because it's working and it's Friday…

IPIGuard: A Novel Tool Dependency Graph-Based Defense Against Indirect Prompt Injection in LLM Agents
Hengyu An, Jinghuai Zhang, Tianyu Du, Chunyi Zhou, Qingming Li, Tao Lin, Shouling Ji
https://arxiv.org/abs/2508.15310

IPIGuard: A Novel Tool Dependency Graph-Based Defense Against Indirect Prompt Injection in LLM Agents
Large language model (LLM) agents are widely deployed in real-world applications, where they leverage tools to retrieve and manipulate external data for complex tasks. However, when interacting with untrusted data sources (e.g., fetching information from public websites), tool responses may contain injected instructions that covertly influence agent behaviors and lead to malicious outcomes, a threat referred to as Indirect Prompt Injection (IPI). Existing defenses typically rely on advanced pro…

There’s a bit in the “Big Man” iPhone film when a character says “your aura is depleted”, which for me was one of those jarring “people vocally saying internet slang” moments.
https://youtu.be/AUrLMJQf2Bw

Collaborative Inference and Learning between Edge SLMs and Cloud LLMs: A Survey of Algorithms, Execution, and Open Challenges
Senyao Li, Haozhao Wang, Wenchao Xu, Rui Zhang, Song Guo, Jingling Yuan, Xian Zhong, Tianwei Zhang, Ruixuan Li
https://arxiv.org/abs/2507.16731

Collaborative Inference and Learning between Edge SLMs and Cloud LLMs: A Survey of Algorithms, Execution, and Open Challenges
As large language models (LLMs) evolve, deploying them solely in the cloud or compressing them for edge devices has become inadequate due to concerns about latency, privacy, cost, and personalization. This survey explores a collaborative paradigm in which cloud-based LLMs and edge-deployed small language models (SLMs) cooperate across both inference and training. We present a unified taxonomy of edge-cloud collaboration strategies. For inference, we categorize approaches into task assignment, t…