Tootfinder

Beyond Log Likelihood: Probability-Based Objectives for Supervised Fine-Tuning across the Model Capability Continuum
Gaotang Li, Ruizhong Qiu, Xiusi Chen, Heng Ji, Hanghang Tong
https://arxiv.org/abs/2510.00526

Beyond Log Likelihood: Probability-Based Objectives for Supervised Fine-Tuning across the Model Capability Continuum
Supervised fine-tuning (SFT) is the standard approach for post-training large language models (LLMs), yet it often shows limited generalization. We trace this limitation to its default training objective: negative log likelihood (NLL). While NLL is classically optimal when training from scratch, post-training operates in a different paradigm and could violate its optimality assumptions, where models already encode task-relevant priors and supervision can be long and noisy. To this end, we study…

Generalizing the Penrose cut-and-paste method: Null shells with pressure and energy flux
Miguel Manzano, Argam Ohanyan, Roland Steinbauer
https://arxiv.org/abs/2508.00231 https:…

Generalizing the Penrose cut-and-paste method: Null shells with pressure and energy flux
The cut-and-paste method is a procedure for constructing null thin shells by matching two regions of the same spacetime across a null hypersurface. Originally proposed by Penrose, it has so far allowed to describe purely gravitational and null-dust shells in constant-curvature backgrounds. In this paper, we extend the cut-and-paste method to null shells with arbitrary gravitational/matter content. To that aim, we first derive a locally Lipschitz continuous form of the metric of the spacetime re…

Beyond Quantification: Navigating Uncertainty in Professional AI Systems
Sylvie Delacroix, Diana Robinson, Umang Bhatt, Jacopo Domenicucci, Jessica Montgomery, Gael Varoquaux, Carl Henrik Ek, Vincent Fortuin, Yulan He, Tom Diethe, Neill Campbell, Mennatallah El-Assady, Soren Hauberg, Ivana Dusparic, Neil Lawrence
https://arxiv.org/abs/2509…

Beyond Quantification: Navigating Uncertainty in Professional AI Systems
The growing integration of large language models across professional domains transforms how experts make critical decisions in healthcare, education, and law. While significant research effort focuses on getting these systems to communicate their outputs with probabilistic measures of reliability, many consequential forms of uncertainty in professional contexts resist such quantification. A physician pondering the appropriateness of documenting possible domestic abuse, a teacher assessing cultu…

kabr-tools: Automated Framework for Multi-Species Behavioral Monitoring
Jenna Kline, Maksim Kholiavchenko, Samuel Stevens, Nina van Tiel, Alison Zhong, Namrata Banerji, Alec Sheets, Sowbaranika Balasubramaniam, Isla Duporge, Matthew Thompson, Elizabeth Campolongo, Jackson Miliko, Neil Rosser, Tanya Berger-Wolf, Charles V. Stewart, Daniel I. Rubenstein
https://

kabr-tools: Automated Framework for Multi-Species Behavioral Monitoring
A comprehensive understanding of animal behavior ecology depends on scalable approaches to quantify and interpret complex, multidimensional behavioral patterns. Traditional field observations are often limited in scope, time-consuming, and labor-intensive, hindering the assessment of behavioral responses across landscapes. To address this, we present kabr-tools (Kenyan Animal Behavior Recognition Tools), an open-source package for automated multi-species behavioral monitoring. This framework in…

Rechter brengt Trumps heffingen forse klap toe. Wat zijn de gevolgen van de uitspraak? | de #Volkskrant
https://www.volkskra…

Priors Matter: Addressing Misspecification in Bayesian Deep Q-Learning
Pascal R. van der Vaart, Neil Yorke-Smith, Matthijs T. J. Spaan
https://arxiv.org/abs/2508.21488 https://

Priors Matter: Addressing Misspecification in Bayesian Deep Q-Learning
Uncertainty quantification in reinforcement learning can greatly improve exploration and robustness. Approximate Bayesian approaches have recently been popularized to quantify uncertainty in model-free algorithms. However, so far the focus has been on improving the accuracy of the posterior approximation, instead of studying the accuracy of the prior and likelihood assumptions underlying the posterior. In this work, we demonstrate that there is a cold posterior effect in Bayesian deep Q-learnin…

A Weighted Regression Approach to Break-Point Detection in Panel Data
Charl Pretorius, Heinrich Roodt
https://arxiv.org/abs/2510.00598 https://arxiv.org/pd…

A Weighted Regression Approach to Break-Point Detection in Panel Data
New procedures for detecting a change in the cross-sectional mean of panel data are proposed. The procedures rely on estimating nuisance parameters using certain cross-sectional means across panels using a weighted least squares regression. In the case of weak cross-sectional dependence between panels, we show how test statistics can be constructed to have a limit null distribution not depending on any choice of bandwidths typically needed to estimate the long-run variances of the panel errors.…

What MLLMs Learn about When they Learn about Multimodal Reasoning: Perception, Reasoning, or their Integration?
Jiwan Chung, Neel Joshi, Pratyusha Sharma, Youngjae Yu, Vibhav Vineet
https://arxiv.org/abs/2510.01719

What MLLMs Learn about When they Learn about Multimodal Reasoning: Perception, Reasoning, or their Integration?
Multimodal reasoning models have recently shown promise on challenging domains such as olympiad-level geometry, yet their evaluation remains dominated by aggregate accuracy, a single score that obscures where and how models are improving. We introduce MathLens, a benchmark designed to disentangle the subskills of multimodal reasoning while preserving the complexity of textbook-style geometry problems. The benchmark separates performance into three components: Perception: extracting information …

Ly-alpha emission reveals two satellite halos around massive groups at z ~ 3: the puzzling case of a quiescent central galaxy
Sicen Guo, Emanuele Daddi, Raphael Gobat, Nikolaj B. Sillassen, Chiara D'Eugenio, R. Michael Rich, Guillaume Elias, Manuel Aravena, Franziska Bruckmann, Camila Correa, Ivan Delvecchio, David Elbaz, Sofia G. Gallego, Fabrizio Gentile, Shuowen Jin, Boris S. Kalita, James D. Neill, Manuel Solimano, Francesco Valentino, Tao Wang

Ly-alpha emission reveals two satellite halos around massive groups at z ~ 3: the puzzling case of a quiescent central galaxy
We present the discovery and characterisation of two Ly$α$ nebulae (LANs), RO-1001-Sat and RO-0959-Sat, as satellite structures of two giant LANs at $z=2.920$ and 3.092. They are found neighbouring two out of four known giant LANs at $z\sim3$ in our MUSE follow-up observations, reinforcing the idea that Ly$α$ emission can be used to trace massive dark matter halos at high-$z$. This high occurrence of massive satellite halos agrees with simulations. With sizes of $\simeq80\times160$ and $80\ti…

Back in an engineering position, I come across newly written #Java code that has a lot of improvement potential regarding maintainability. It works, but I imagine engineers having to update it, including the original author’s future self and me, and I’m sure I can help them. This week, I refactored code using creational