Tootfinder

SP-MoE: Speculative Decoding and Prefetching for Accelerating MoE-based Model Inference
Liangkun Chen, Zijian Wen, Tian Wu, Xiaoxi Zhang, Chuan Wu
https://arxiv.org/abs/2510.10302

SP-MoE: Speculative Decoding and Prefetching for Accelerating MoE-based Model Inference
The Mixture-of-Experts (MoE) architecture has been widely adopted in large language models (LLMs) to reduce computation cost through model sparsity. Employing speculative decoding (SD) can further accelerate MoE inference by drafting multiple tokens per step and verifying them in parallel. However, combining MoE with SD inflates GPU memory and aggravates CPU-GPU bandwidth contention during multi-token verification. Existing MoE offloading systems are SD-agnostic and do not address this bottlene…

FuseCodec: Semantic-Contextual Fusion and Supervision for Neural Codecs
Md Mubtasim Ahasan, Rafat Hasan Khan, Tasnim Mohiuddin, Aman Chadha, Tariq Iqbal, M Ashraful Amin, Amin Ahsan Ali, Md Mofijul Islam, A K M Mahbubur Rahman
https://arxiv.org/abs/2509.11425

FuseCodec: Semantic-Contextual Fusion and Supervision for Neural Codecs
Speech tokenization enables discrete representation and facilitates speech language modeling. However, existing neural codecs capture low-level acoustic features, overlooking the semantic and contextual cues inherent to human speech. While recent efforts introduced semantic representations from self-supervised speech models or incorporated contextual representations from pre-trained language models, challenges remain in aligning and unifying the semantic and contextual representations. We intro…

Inductive Bias Extraction and Matching for LLM Prompts
Christian M. Angel, Francis Ferraro
https://arxiv.org/abs/2508.10295 https://arxiv.org/pdf/2508.1029…

Inductive Bias Extraction and Matching for LLM Prompts
The active research topic of prompt engineering makes it evident that LLMs are sensitive to small changes in prompt wording. A portion of this can be ascribed to the inductive bias that is present in the LLM. By using an LLM's output as a portion of its prompt, we can more easily create satisfactory wording for prompts. This has the effect of creating a prompt that matches the inductive bias in model. Empirically, we show that using this Inductive Bias Extraction and Matching strategy improves …

A Computation of Tamarkin-Tsygan Calculus
Jun Chen, Xiabing Ruan, Jia Yang
https://arxiv.org/abs/2509.12984 https://arxiv.org/pdf/2509.12984

A Computation of Tamarkin-Tsygan Calculus
We compute the full Tamarkin-Tsygan calculus of a Koszul algebra whose global dimension exceeds the number of generators. Our results show that even for algebras possessing an economic presentation and agreeable homological properties, the Hochschild (co)homology, as well as the structure of the Tamarkin--Tsygan calculus may exhibit a rather intricate behavior.

Been to the hospital and had the cast removed.
Should make typing easier.
More pain now than when I first went there though. RSI aggravated by the cast. Thumb fucked up by the cast. Wrist hurts when moved.
Apparently I'm to stretch the wrist slowly and gradually recover mobility like wrist yoga. Physio to come maybe.
Have booked a massage for tomorrow to try and realign my frame, all fucked up by the cast and the sitting around for a month.
Do not recommend fracturing bones in your arm. It's an bad scene man.

MATRIX: Multimodal Agent Tuning for Robust Tool-Use Reasoning
Tajamul Ashraf, Umair Nawaz, Abdelrahman M. Shaker, Rao Anwer, Philip Torr, Fahad Shahbaz Khan, Salman Khan
https://arxiv.org/abs/2510.08567

MATRIX: Multimodal Agent Tuning for Robust Tool-Use Reasoning
Vision language models (VLMs) are increasingly deployed as controllers with access to external tools for complex reasoning and decision-making, yet their effectiveness remains limited by the scarcity of high-quality multimodal trajectories and the cost of manual annotation. We address this challenge with a vision-centric agent tuning framework that automatically synthesizes multimodal trajectories, generates step-wise preference pairs, and trains a VLM controller for robust tool-use reasoning. …

Shout-out an Münsteraner Studierende: Das Münster Center for Open Science (MüCOS) sucht studentische Hilfskräfte!
Bei Interesse beim @… melden!
https://www.uni-muenster.de/MueCOS/

The Underground Muon Detector of AugerPrime: Status and Performance
Joaquin de Jesus (for the Pierre Auger Collaboration)
https://arxiv.org/abs/2509.08631 https://

The Underground Muon Detector of AugerPrime: Status and Performance
As part of the upgrade of the Pierre Auger Observatory, known as AugerPrime, the Underground Muon Detector is being installed in the low-energy extension of the Surface Detector, allowing for a direct measurement of the muonic component of air showers produced by ultra-high-energy cosmic rays with energies between 10^{16.8} and 10^{19} eV. The detector consists of an array of 30 m2 plastic scintillator detectors buried 2.3 m underground near a water-Cherenkov detector. Two modes of operation ar…

Leveraging large language models for SQL behavior-based database intrusion detection
Meital Shlezinger, Shay Akirav, Lei Zhou, Liang Guo, Avi Kessel, Guoliang Li
https://arxiv.org/abs/2508.05690

Leveraging large language models for SQL behavior-based database intrusion detection
Database systems are extensively used to store critical data across various domains. However, the frequency of abnormal database access behaviors, such as database intrusion by internal and external attacks, continues to rise. Internal masqueraders often have greater organizational knowledge, making it easier to mimic employee behavior effectively. In contrast, external masqueraders may behave differently due to their lack of familiarity with the organization. Current approaches lack the granul…

The simulation chain for the Terzina Cherenkov telescope on board the NUSES space mission
M. Abdullahi, R. Aloisio, F. Arneodo, S. Ashurov, U. Atalay, F. C. T. Barbato, R. Battiston, M. Bertaina, E. Bissaldi, D. Boncioli, L. Burmistrov, F. Cadoux, I. Cagnoli, E. Casilli, D. Cortis, A. Cummings, M. D'Arco, S. Davarpanah, I. De Mitri, G. De Robertis, A. Di Giovanni, A. Di Salvo, L. Di Venere, J. Eser, Y. Favre, S. Fogliacco, G. Fontanella, P. Fusco, S. Garbolino, F. Gargano, M. Gilib…

The simulation chain for the Terzina Cherenkov telescope on board the NUSES space mission
The Terzina telescope is designed to detect ultra-high energy cosmic rays (UHECRs) and Earth-skimming neutrinos from a 550 km low-Earth orbit (LEO) by observing Cherenkov light emitted by Extensive Air Showers (EAS) in the Earth's atmosphere pointing towards the telescope and in the field of view. In this contribution, a simulation chain for the Terzina telescope on board the NUSES mission will be presented. The chain encompasses all stages of the detection process, from event generation and EA…