Operational forecasting of the ionosphere remains a critical space weather challenge due to sparse observations, complex coupling across geospatial layers, and a growing need for timely, accurate predictions that support Global Navigation Satellite System (GNSS), communications, aviation safety, as well as satellite operations. As part of the 2025 NASA Heliolab, we present a curated, open-access dataset that integrates diverse ionospheric and heliospheric measurements into a coherent, machine learning-ready structure, designed specifically to support next-generation forecasting models and address gaps in current operational frameworks. Our workflow integrates a large selection of data sources comprising Solar Dynamic Observatory data, solar irradiance indices (F10.7), solar wind parameters (velocity and interplanetary magnetic field), geomagnetic activity indices (Kp, AE, SYM-H), and NASA JPL's Global Ionospheric Maps of Total Electron Content (GIM-TEC). We also implement geospatially sparse data such as the TEC derived from the World-Wide GNSS Receiver Network and crowdsourced Android smartphone measurements. This novel heterogeneous dataset is temporally and spatially aligned into a single, modular data structure that supports both physical and data-driven modeling. Leveraging this dataset, we train and benchmark several spatiotemporal machine learning architectures for forecasting vertical TEC under both quiet and geomagnetically active conditions. This work presents an extensive dataset and modeling pipeline that enables exploration of not only ionospheric dynamics but also broader Sun-Earth interactions, supporting both scientific inquiry and operational forecasting efforts.

One way to address safety risks from large language models (LLMs) is to censor dangerous knowledge from their training data. While this removes the explicit information, implicit information can remain scattered across various training documents. Could an LLM infer the censored knowledge by piecing together these implicit hints? As a step towards answering this question, we study inductive out-of-context reasoning (OOCR), a type of generalization in which LLMs infer latent information from evidence distributed across training documents and apply it to downstream tasks without in-context learning. Using a suite of five tasks, we demonstrate that frontier LLMs can perform inductive OOCR.

Large language models (LLMs) frequently generate hallucinations-content that deviates from factual accuracy or provided context-posing challenges for diagnosis due to the complex interplay of underlying causes. This paper introduces a subsequence association framework to systematically trace and understand hallucinations. Our key insight is that hallucinations arise when dominant hallucinatory associations outweigh faithful ones. Through theoretical and empirical analyses, we demonstrate that decoder-only transformers effectively function as subsequence embedding models, with linear layers encoding input-output associations. We propose a tracing algorithm that identifies causal subsequences by analyzing hallucination probabilities across randomized input contexts. Experiments show our method outperforms standard attribution techniques in identifying hallucination causes and aligns with evidence from the model's training corpus. This work provides a unified perspective on hallucinations and a robust framework for their tracing and analysis.

Track reconstruction is a vital aspect of High-Energy Physics (HEP) and plays a critical role in major experiments. In this study, we delve into unexplored avenues for particle track reconstruction and hit clustering. Firstly, we enhance the algorithmic design effort by utilising a simplified simulator (REDVID) to generate training data that is specifically composed for simplicity. We demonstrate the effectiveness of this data in guiding the development of optimal network architectures. Additionally, we investigate the application of image segmentation networks for this task, exploring their potential for accurate track reconstruction. Moreover, we approach the task from a different perspective by treating it as a hit sequence to track sequence translation problem. Specifically, we explore the utilisation of Transformer architectures for tracking purposes. Our preliminary findings are covered in detail. By considering this novel approach, we aim to uncover new insights and potential advancements in track reconstruction. This research sheds light on previously unexplored methods and provides valuable insights for the field of particle track reconstruction and hit clustering in HEP.

Particle tracking is crucial for almost all physics analysis programs at the Large Hadron Collider. Deep learning models are pervasively used in particle tracking related tasks. However, the current practice is to design and train one deep learning model for one task with supervised learning techniques. The trained models work well for tasks they are trained on but show no or little generalization capabilities. We propose to unify these models with a language model. In this paper, we present a tokenized detector representation that allows us to train a BERT model for particle tracking. The trained BERT model, namely TrackingBERT, offers latent detector module embedding that can be used for other tasks. This work represents the first step towards developing a foundational model for particle detector understanding.

With the emergence of pretrained vision-language models (VLMs), considerable efforts have been devoted to fine-tuning them for downstream tasks. Despite the progress made in designing efficient fine-tuning methods, such methods require access to the model's parameters, which can be challenging as model owners often opt to provide their models as a black box to safeguard model ownership. This paper proposes a \textbf{C}ollabo\textbf{ra}tive \textbf{F}ine-\textbf{T}uning (\textbf{CraFT}) approach for fine-tuning black-box VLMs to downstream tasks, where one only has access to the input prompts and the output predictions of the model. CraFT comprises two modules, a prompt generation module for learning text prompts and a prediction refinement module for enhancing output predictions in residual style. Additionally, we introduce an auxiliary prediction-consistent loss to promote consistent optimization across these modules. These modules are optimized by a novel collaborative training algorithm. Extensive experiments on few-shot classification over 15 datasets demonstrate the superiority of CraFT. The results show that CraFT achieves a decent gain of about 12\% with 16-shot datasets and only 8,000 queries. Moreover, CraFT trains faster and uses only about 1/80 of the memory footprint for deployment, while sacrificing only 1.62\% compared to the white-box method.

A major challenge in sample-based inference (SBI) for Bayesian neural networks is the size and structure of the networks' parameter space. Our work shows that successful SBI is possible by embracing the characteristic relationship between weight and function space, uncovering a systematic link between overparameterization and the difficulty of the sampling problem. Through extensive experiments, we establish practical guidelines for sampling and convergence diagnosis. As a result, we present a Bayesian deep ensemble approach as an effective solution with competitive performance and uncertainty quantification.

The application of Graph Neural Networks (GNN) in track reconstruction is a promising approach to cope with the challenges arising at the High-Luminosity upgrade of the Large Hadron Collider (HL-LHC). GNNs show good track-finding performance in high-multiplicity scenarios and are naturally parallelizable on heterogeneous compute architectures. Typical high-energy-physics detectors have high resolution in the innermost layers to support vertex reconstruction but lower resolution in the outer parts. GNNs mainly rely on 3D space-point information, which can cause reduced track-finding performance in the outer regions. In this contribution, we present a novel combination of GNN-based track finding with the classical Combinatorial Kalman Filter (CKF) algorithm to circumvent this issue: The GNN resolves the track candidates in the inner pixel region, where 3D space points can represent measurements very well. These candidates are then picked up by the CKF in the outer regions, where the CKF performs well even for 1D measurements. Using the ACTS infrastructure, we present a proof of concept based on truth tracking in the pixels as well as a dedicated GNN pipeline trained on $t\bar{t}$ events with pile-up 200 in the OpenDataDetector.

Traditional charged particle tracking algorithms at the Large Hadron Collider (LHC) are based on the combinatorial Kalman filter. However, this class of algorithms exhibits sub-optimal scaling with respect to pileup, rendering tracking a bottleneck for future experiments such as the High Luminosity LHC (HL-LHC) [1]. This has prompted research into tracking algorithms leveraging graph neural networks (GNNs) or similar machine learning (ML) architectures demonstrating improved computational scaling. Recent results have confirmed that GNN-based algorithms can indeed achieve linear scaling with pileup [2, 3]. The majority of GNN approaches adopt an edge classification (EC) approach to tackle the tracking problem. Given an initial graph that connects all hits that potentially belong to the same particle, a GNN is trained to remove edges that connect hits belonging to different particles.

Providing consistent, individualized feedback to teachers on their instruction can improve student learning outcomes. Such feedback can especially benefit novice instructors who teach on online platforms and have limited access to instructional training. To build scalable measures of instruction, we fine-tune RoBERTa and GPT models to identify five instructional talk moves inspired by accountable talk theory: adding on, connecting, eliciting, probing and revoicing students' ideas. We fine-tune these models on a newly annotated dataset of 2500 instructor utterances derived from transcripts of small group instruction in an online computer science course, Code in Place. Although we find that GPT-3 consistently outperforms RoBERTa in terms of precision, its recall varies significantly. We correlate the instructors' use of each talk move with indicators of student engagement and satisfaction, including students' section attendance, section ratings, and assignment completion rates. We find that using talk moves generally correlates positively with student outcomes, and connecting student ideas has the largest positive impact. These results corroborate previous research on the effectiveness of accountable talk moves and provide exciting avenues for using these models to provide instructors with useful, scalable feedback.