Search

The practical use of text-to-image generation has evolved from simple, monolithic models to complex workflows that combine multiple specialized components. While workflow-based approaches can lead to improved image quality, crafting effective workflows requires significant expertise, owing to the large number of available components, their complex inter-dependence, and their dependence on the generation prompt. Here, we introduce the novel task of prompt-adaptive workflow generation, where the goal is to automatically tailor a workflow to each user prompt. We propose two LLM-based approaches to tackle this task: a tuning-based method that learns from user-preference data, and a training-free method that uses the LLM to select existing flows. Both approaches lead to improved image quality when compared to monolithic models or generic, prompt-independent workflows. Our work shows that prompt-dependent flow prediction offers a new pathway to improving text-to-image generation quality, complementing existing research directions in the field., Comment: Project website: https://comfygen-paper.github.io/

The KM3NeT Collaboration has tackled a common challenge faced by the astroparticle physics community, namely adapting the experiment-specific simulation software to work with the CORSIKA air shower simulation output. The proposed solution is an extension of the open-source code gSeaGen, allowing for the transport of muons generated by CORSIKA to a detector of any size at an arbitrary depth. The gSeaGen code was not only extended in terms of functionalities but also underwent a thorough redesign of the muon propagation routine, resulting in a more accurate and efficient simulation. This paper presents the capabilities of the new gSeaGen code as well as prospects for further developments., Comment: 27 pages, 13 figures

We study the implicit bias of the general family of steepest descent algorithms, which includes gradient descent, sign descent and coordinate descent, in deep homogeneous neural networks. We prove that an algorithm-dependent geometric margin starts increasing once the networks reach perfect training accuracy and characterize the late-stage bias of the algorithms. In particular, we define a generalized notion of stationarity for optimization problems and show that the algorithms progressively reduce a (generalized) Bregman divergence, which quantifies proximity to such stationary points of a margin-maximization problem. We then experimentally zoom into the trajectories of neural networks optimized with various steepest descent algorithms, highlighting connections to the implicit bias of Adam.

We here present an improved version of the Sortition Foundation's GROUPSELECT software package, which aims to repeatedly allocate participants of a deliberative process to discussion groups in a way that balances demographics in each group and maximises distinct meetings over time. Our result, GROUPOPT, significantly outperforms the prior algorithmic approach LEGACY. We also add functionalities to the GROUPSELECT software to help the end user. The GROUPOPT algorithm utilises random shuffles and Pareto swaps to find a locally optimal solution that maximises demographic balance and minimises the number of pairwise previous meetings, with the relative importance of these two metrics defined by the user., Comment: 13 pages, 3 figures

Finding meaningful distances between high-dimensional data samples is an important scientific task. To this end, we propose a new tree-Wasserstein distance (TWD) for high-dimensional data with two key aspects. First, our TWD is specifically designed for data with a latent feature hierarchy, i.e., the features lie in a hierarchical space, in contrast to the usual focus on embedding samples in hyperbolic space. Second, while the conventional use of TWD is to speed up the computation of the Wasserstein distance, we use its inherent tree as a means to learn the latent feature hierarchy. The key idea of our method is to embed the features into a multi-scale hyperbolic space using diffusion geometry and then present a new tree decoding method by establishing analogies between the hyperbolic embedding and trees. We show that our TWD computed based on data observations provably recovers the TWD defined with the latent feature hierarchy and that its computation is efficient and scalable. We showcase the usefulness of the proposed TWD in applications to word-document and single-cell RNA-sequencing datasets, demonstrating its advantages over existing TWDs and methods based on pre-trained models.

This exploratory pilot study investigated the potential of combining a domain-specific model, BERN2, with large language models (LLMs) to enhance automated disease phenotyping from research survey data. Motivated by the need for efficient and accurate methods to harmonize the growing volume of survey data with standardized disease ontologies, we employed BERN2, a biomedical named entity recognition and normalization model, to extract disease information from the ORIGINS birth cohort survey data. After rigorously evaluating BERN2's performance against a manually curated ground truth dataset, we integrated various LLMs using prompt engineering, Retrieval-Augmented Generation (RAG), and Instructional Fine-Tuning (IFT) to refine the model's outputs. BERN2 demonstrated high performance in extracting and normalizing disease mentions, and the integration of LLMs, particularly with Few Shot Inference and RAG orchestration, further improved accuracy. This approach, especially when incorporating structured examples, logical reasoning prompts, and detailed context, offers a promising avenue for developing tools to enable efficient cohort profiling and data harmonization across large, heterogeneous research datasets.

Low back pain is the symptom that is the second most frequently reported to primary care physicians, effecting 50 to 80 percent of the population in a lifetime, resulting in multiple referrals of patients suffering from back problems, to CT and MRI scans, which are then examined by radiologists. The radiologists examining these spinal scans naturally focus on spinal pathologies and might miss other types of abnormalities, and in particular, abdominal ones, such as malignancies. Nevertheless, the patients whose spine was scanned might as well have malignant and other abdominal pathologies. Thus, clinicians have suggested the need for computerized assistance and decision support in screening spinal scans for additional abnormalities. In the current study, We have addressed the important case of detecting suspicious lesions in the adrenal glands as an example for the overall methodology we have developed. A patient CT scan is integrated from multiple slices with an axial orientation. Our method determines whether a patient has an abnormal adrenal gland, and localises the abnormality if it exists. Our method is composed of three deep learning models; each model has a different task for achieving the final goal. We call our compound method the Multi Model Graph Aggregation MMGA method. The novelty in this study is twofold. First, the use, for an important screening task, of CT scans that are originally focused and tuned for imaging the spine, which were acquired from patients with potential spinal disorders, for detection of a totally different set of abnormalities such as abdominal Adrenal glands pathologies. Second, we have built a complex pipeline architecture composed from three deep learning models that can be utilized for other organs (such as the pancreas or the kidney), or for similar applications, but using other types of imaging, such as MRI.

Lately, there has been a surge in interest surrounding generative modeling of time series data. Most existing approaches are designed either to process short sequences or to handle long-range sequences. This dichotomy can be attributed to gradient issues with recurrent networks, computational costs associated with transformers, and limited expressiveness of state space models. Towards a unified generative model for varying-length time series, we propose in this work to transform sequences into images. By employing invertible transforms such as the delay embedding and the short-time Fourier transform, we unlock three main advantages: i) We can exploit advanced diffusion vision models; ii) We can remarkably process short- and long-range inputs within the same framework; and iii) We can harness recent and established tools proposed in the time series to image literature. We validate the effectiveness of our method through a comprehensive evaluation across multiple tasks, including unconditional generation, interpolation, and extrapolation. We show that our approach achieves consistently state-of-the-art results against strong baselines. In the unconditional generation tasks, we show remarkable mean improvements of 58.17% over previous diffusion models in the short discriminative score and 132.61% in the (ultra-)long classification scores. Code is at https://github.com/azencot-group/ImagenTime., Comment: Accepted to NeurIPS 2024; The first two authors contributed equally

We study the overfitting behavior of fully connected deep Neural Networks (NNs) with binary weights fitted to perfectly classify a noisy training set. We consider interpolation using both the smallest NN (having the minimal number of weights) and a random interpolating NN. For both learning rules, we prove overfitting is tempered. Our analysis rests on a new bound on the size of a threshold circuit consistent with a partial function. To the best of our knowledge, ours are the first theoretical results on benign or tempered overfitting that: (1) apply to deep NNs, and (2) do not require a very high or very low input dimension., Comment: 60 pages, 4 figures

Large Language Models (LLMs) are known to hallucinate, whereby they generate plausible but inaccurate text. This phenomenon poses significant risks in critical applications, such as medicine or law, necessitating robust hallucination mitigation strategies. While recent works have proposed fine-tuning methods to teach LLMs to abstain from answering questions beyond their knowledge or capabilities, these methods rely on the existence of ground-truth labels or are limited to short-form responses. To address these limitations, we propose fine-tuning using semantic entropy, an uncertainty measure derived from introspection into the model which does not require external labels. We demonstrate that our approach matches or outperforms models fine-tuned using prior work and achieves strong performance for both short and long-form generations on a range of datasets., Comment: Accepted to NeurIPS Safe Generative AI Workshop 2024

Scaling inference compute in large language models (LLMs) through repeated sampling consistently increases the coverage (fraction of problems solved) as the number of samples increases. We conjecture that this observed improvement is partially due to the answer distribution of standard evaluation benchmarks, which is skewed towards a relatively small set of common answers. To test this conjecture, we define a baseline that enumerates answers according to their prevalence in the training set. Experiments spanning two domains -- mathematical reasoning and factual knowledge -- reveal that this baseline outperforms repeated model sampling for some LLMs, while the coverage for others is on par with that of a mixture strategy that obtains $k$ answers by using only $10$ model samples and similarly guessing the remaining $k-10$ attempts via enumeration. Our baseline enables a more accurate measurement of how much repeated sampling improves coverage in such settings beyond prompt-agnostic guessing.

Measurements of the EMC effect in the tritium and helium-3 mirror nuclei are reported. The data were obtained by the MARATHON Jefferson Lab experiment, which performed deep inelastic electron scattering from deuterium and the three-body nuclei, using a cryogenic gas target system and the High Resolution Spectrometers of the Hall A Facility of the Lab. The data cover the Bjorken $x$ range from 0.20 to 0.83, corresponding to a squared four-momentum transfer $Q^2$ range from 2.7 to $11.9\gevsq$, and to an invariant mass $W$ of the final hadronic state greater than 1.84 GeV/${\it c}^2$. The tritium EMC effect measurement is the first of its kind. The MARATHON experimental results are compared to results from previous measurements by DESY-HERMES and JLab-Hall C experiments, as well as with few-body theoretical predictions., Comment: arXiv admin note: text overlap with arXiv:2104.05850

The emergence of neural representations has revolutionized our means for digitally viewing a wide range of 3D scenes, enabling the synthesis of photorealistic images rendered from novel views. Recently, several techniques have been proposed for connecting these low-level representations with the high-level semantics understanding embodied within the scene. These methods elevate the rich semantic understanding from 2D imagery to 3D representations, distilling high-dimensional spatial features onto 3D space. In our work, we are interested in connecting language with a dynamic modeling of the world. We show how to lift spatio-temporal features to a 4D representation based on 3D Gaussian Splatting. This enables an interactive interface where the user can spatiotemporally localize events in the video from text prompts. We demonstrate our system on public 3D video datasets of people and animals performing various actions., Comment: Project webpage: https://tau-vailab.github.io/4-LEGS/

The robustness of LLMs to jailbreak attacks, where users design prompts to circumvent safety measures and misuse model capabilities, has been studied primarily for LLMs acting as simple chatbots. Meanwhile, LLM agents -- which use external tools and can execute multi-stage tasks -- may pose a greater risk if misused, but their robustness remains underexplored. To facilitate research on LLM agent misuse, we propose a new benchmark called AgentHarm. The benchmark includes a diverse set of 110 explicitly malicious agent tasks (440 with augmentations), covering 11 harm categories including fraud, cybercrime, and harassment. In addition to measuring whether models refuse harmful agentic requests, scoring well on AgentHarm requires jailbroken agents to maintain their capabilities following an attack to complete a multi-step task. We evaluate a range of leading LLMs, and find (1) leading LLMs are surprisingly compliant with malicious agent requests without jailbreaking, (2) simple universal jailbreak templates can be adapted to effectively jailbreak agents, and (3) these jailbreaks enable coherent and malicious multi-step agent behavior and retain model capabilities. To enable simple and reliable evaluation of attacks and defenses for LLM-based agents, we publicly release AgentHarm at https://huggingface.co/datasets/ai-safety-institute/AgentHarm.

A crucial capability of Machine Learning models in real-world applications is the ability to continuously learn new tasks. This adaptability allows them to respond to potentially inevitable shifts in the data-generating distribution over time. However, in Continual Learning (CL) settings, models often struggle to balance learning new tasks (plasticity) with retaining previous knowledge (memory stability). Consequently, they are susceptible to Catastrophic Forgetting, which degrades performance and undermines the reliability of deployed systems. Variational Continual Learning methods tackle this challenge by employing a learning objective that recursively updates the posterior distribution and enforces it to stay close to the latest posterior estimate. Nonetheless, we argue that these methods may be ineffective due to compounding approximation errors over successive recursions. To mitigate this, we propose new learning objectives that integrate the regularization effects of multiple previous posterior estimations, preventing individual errors from dominating future posterior updates and compounding over time. We reveal insightful connections between these objectives and Temporal-Difference methods, a popular learning mechanism in Reinforcement Learning and Neuroscience. We evaluate the proposed objectives on challenging versions of popular CL benchmarks, demonstrating that they outperform standard Variational CL methods and non-variational baselines, effectively alleviating Catastrophic Forgetting.

The phenomenon of benign overfitting, where a trained neural network perfectly fits noisy training data but still achieves near-optimal test performance, has been extensively studied in recent years for linear models and fully-connected/convolutional networks. In this work, we study benign overfitting in a single-head softmax attention model, which is the fundamental building block of Transformers. We prove that under appropriate conditions, the model exhibits benign overfitting in a classification setting already after two steps of gradient descent. Moreover, we show conditions where a minimum-norm/maximum-margin interpolator exhibits benign overfitting. We study how the overfitting behavior depends on the signal-to-noise ratio (SNR) of the data distribution, namely, the ratio between norms of signal and noise tokens, and prove that a sufficiently large SNR is both necessary and sufficient for benign overfitting.

We study what provable privacy attacks can be shown on trained, 2-layer ReLU neural networks. We explore two types of attacks; data reconstruction attacks, and membership inference attacks. We prove that theoretical results on the implicit bias of 2-layer neural networks can be used to provably reconstruct a set of which at least a constant fraction are training points in a univariate setting, and can also be used to identify with high probability whether a given point was used in the training set in a high dimensional setting. To the best of our knowledge, our work is the first to show provable vulnerabilities in this setting.

(Abridged) We explore the chemistry of the most abundant C, O, S, and N bearing species in molecular clouds, in the context of the IRAM 30 m Large Programme Gas phase Elemental abundances in Molecular Clouds (GEMS). In this work, we aim to assess the limitations introduced in the observational works when a uniform density is assumed along the line of sight for fitting the observations, developing a very simple numerical model of a turbulent box. We perform a MHD simulation in order to reproduce the turbulent steady-state of a turbulent box with properties typical of a molecular filament before collapse. We post-process the results of the MHD simulation with a chemical code to predict molecular abundances, and then post-process this cube with a radiative transfer code to create synthetic emission maps for a series of rotational transitions observed during the GEMS project. From the chemical point of view, we find that turbulence produces variations on the predicted abundances, but they are more or less critical depending on the chosen transition and the chemical age. When compared to real observations, the results from the turbulent simulation provides a better fit than when assuming a uniform gas distribution along the line of sight. In the view of our results, we conclude that taking into account turbulence when fitting observations might significantly improve the agreement with model predictions. This is especially important for sulfur bearing species that are very sensitive to the variations of density produced by turbulence at early times (0.1 Myr). The abundance of CO is also quite sensitive to turbulence when considering the evolution beyond a few 0.1 Myr., Comment: Published in Astronomy and Astrophysics

This paper introduces Idempotent Test-Time Training (IT$^3$), a novel approach to addressing the challenge of distribution shift. While supervised-learning methods assume matching train and test distributions, this is rarely the case for machine learning systems deployed in the real world. Test-Time Training (TTT) approaches address this by adapting models during inference, but they are limited by a domain specific auxiliary task. IT$^3$ is based on the universal property of idempotence. An idempotent operator is one that can be applied sequentially without changing the result beyond the initial application, that is $f(f(x))=f(x)$. At training, the model receives an input $x$ along with another signal that can either be the ground truth label $y$ or a neutral "don't know" signal $0$. At test time, the additional signal can only be $0$. When sequentially applying the model, first predicting $y_0 = f(x, 0)$ and then $y_1 = f(x, y_0)$, the distance between $y_0$ and $y_1$ measures certainty and indicates out-of-distribution input $x$ if high. We use this distance, that can be expressed as $||f(x, f(x, 0)) - f(x, 0)||$ as our TTT loss during inference. By carefully optimizing this objective, we effectively train $f(x,\cdot)$ to be idempotent, projecting the internal representation of the input onto the training distribution. We demonstrate the versatility of our approach across various tasks, including corrupted image classification, aerodynamic predictions, tabular data with missing information, age prediction from face, and large-scale aerial photo segmentation. Moreover, these tasks span different architectures such as MLPs, CNNs, and GNNs.

We introduce a novel platform for realizing interaction-induced Hall crystals with diverse Chern numbers $C$. This platform consists of a two-dimensional semiconductor or graphene subjected to an out-of-plane magnetic field and a one-dimensional modulation, which can be realized by moir\'e or dielectric engineering. We show that interactions drive the system to spontaneously break the residual translational symmetry, resulting in Hall crystals with various $C$ (including $|C|>1$). Remarkably, these phases persist across continuous ranges of filling and magnetic field, and the global phase diagram can be understood in a unified manner., Comment: 7+7 pages

Transformers have the capacity to act as supervised learning algorithms: by properly encoding a set of labeled training ("in-context") examples and an unlabeled test example into an input sequence of vectors of the same dimension, the forward pass of the transformer can produce predictions for that unlabeled test example. A line of recent work has shown that when linear transformers are pre-trained on random instances for linear regression tasks, these trained transformers make predictions using an algorithm similar to that of ordinary least squares. In this work, we investigate the behavior of linear transformers trained on random linear classification tasks. Via an analysis of the implicit regularization of gradient descent, we characterize how many pre-training tasks and in-context examples are needed for the trained transformer to generalize well at test-time. We further show that in some settings, these trained transformers can exhibit "benign overfitting in-context": when in-context examples are corrupted by label flipping noise, the transformer memorizes all of its in-context examples (including those with noisy labels) yet still generalizes near-optimally for clean test examples., Comment: 34 pages

Neutrinos described as an open quantum system may interact with the environment which introduces stochastic perturbations to their quantum phase. This mechanism leads to a loss of coherence along the propagation of the neutrino $-$ a phenomenon commonly referred to as decoherence $-$ and ultimately, to a modification of the oscillation probabilities. Fluctuations in space-time, as envisaged by various theories of quantum gravity, are a potential candidate for a decoherence-inducing environment. Consequently, the search for decoherence provides a rare opportunity to investigate quantum gravitational effects which are usually beyond the reach of current experiments. In this work, quantum decoherence effects are searched for in neutrino data collected by the KM3NeT/ORCA detector from January 2020 to November 2021. The analysis focuses on atmospheric neutrinos within the energy range of a few GeV to $100\,\mathrm{GeV}$. Adopting the open quantum system framework, decoherence is described in a phenomenological manner with the strength of the effect given by the parameters $\Gamma_{21}$ and $\Gamma_{31}$. Following previous studies, a dependence of the type $\Gamma_{ij} \propto (E/E_0)^n$ on the neutrino energy is assumed and the cases $n = -2,-1$ are explored. No significant deviation with respect to the standard oscillation hypothesis is observed. Therefore, $90\,\%$ CL upper limits are estimated as $\Gamma_{21} < 4.6\cdot 10^{-21}\,$GeV and $\Gamma_{31} < 8.4\cdot 10^{-21}\,$GeV for $n = -2$, and $\Gamma_{21} < 1.9\cdot 10^{-22}\,$GeV and $\Gamma_{31} < 2.7\cdot 10^{-22}\,$GeV for $n = -1$, respectively., Comment: 17 pages, 5 figures

The Charting Cluster Construction with VUDS and ORELSE (C3VO) survey is an ongoing imaging and spectroscopic campaign aiming to map out the growth of structure up to $z\sim5$ and was born from the combination of the VIMOS Ultra Deep Survey (VUDS) and the Observations of Redshift Evolution in Large-Scale Environments (ORELSE) survey. As we previously accomplished with the ORELSE survey, we apply our technique known as Voronoi tessellation Monte-Carlo (VMC) mapping to search for serendipitous galaxy overdensities at $2 14$. In the C3VO fields, we were able to recover many of the previously known structures in the literature as well as find hundreds of new overdensity candidates, once again demonstrating the powerful capabilities of VMC mapping when applied to wide-field optical and infrared galaxy evolution surveys at ever higher redshifts., Comment: 36 pages, 16 figures, 5 tables, submitted to ApJ

The open-ended coaxial probe (OECP) technique is one of the most commonly used methods for the characterization of homogeneous media properties, especially in the biomedical sciences. However, when considering inhomogeneous media, the effect of the heterogeneity on the probe terminal admittance is unclear, making the measured admittance hard to interpret and relate to the medium properties. In this paper we present an analytical model for the contribution of an isotropic scatterer embedded in an otherwise homogeneous medium to the probe admittance. We utilize rigorous scattering theory and various approximations to obtain simplified, closed-form expressions. Using the obtained results we present a method to accurately extract the scatterer properties from a measurement of the admittance. In addition, we define the sensing depth, and show how it can be mapped as a function of the expected scatterer properties. Full-wave simulations are used to verify the analytical model, and the proposed method paves a path for further generalization to additional scenarios of open-coaxial probe sensing of an inhomogeneous medium., Comment: 7 pages, 6 figures

Detections of fast X-ray transients (FXTs) have been accrued over the last few decades. However, their origin has remained mysterious. There is now rapid progress thanks to timely discoveries and localisations with the Einstein Probe mission. Early results indicate that FXTs may frequently, but not always, be associated with gamma-ray bursts (GRBs). Here, we report on the multi-wavelength counterpart of FXT EP240414a, which has no reported gamma-ray counterpart. The transient is located 25.7~kpc in projection from a massive galaxy at $z=0.40$. We perform comprehensive photometric and spectroscopic follow-up. The optical light curve shows at least three distinct emission episodes with timescales of $\sim 1, 4$ and 15 days and peak absolute magnitudes of $M_R \sim -20$, $-21$, and $-19.5$, respectively. The optical spectrum at early times is extremely blue, inconsistent with afterglow emission. It may arise from the interaction of both jet and supernova shock waves with the stellar envelope and a dense circumstellar medium, as has been suggested for some Fast Blue Optical Transients (LFBOTs). At late times, the spectrum evolves to a broad-lined~Type~Ic supernova, similar to those seen in collapsar long-GRBs. This implies that the progenitor of EP240414a is a massive star creating a jet-forming supernova inside a dense envelope, resulting in an X-ray outburst with a luminosity of $\sim 10^{48}$ erg s$^{-1}$, and the complex observed optical/IR light curves. If correct, this argues for a causal link between the progenitors of long-GRBs, FXTs and LFBOTs., Comment: 36 pages, 13 figures, submitted to ApJ

We present photometric and spectroscopic observations of SN2020xga and SN2022xgc, two hydrogen-poor superluminous supernovae (SLSNe-I) at $z = 0.4296$ and $z = 0.3103$ respectively, that show an additional set of broad Mg II absorption lines, blueshifted by a few thousand km s$^{-1}$ with respect to the host galaxy absorption system. Previous work interpreted this as due to resonance line scattering of the SLSN continuum by rapidly expanding CSM expelled shortly before the explosion. The peak rest-frame $g$-band magnitude of SN2020xga is $-22.30 \pm 0.04$ mag and of SN2022xgc is $-21.97 \pm 0.05$ mag, placing them among the brightest SLSNe-I. We use high-quality spectra from ultraviolet to near-infrared wavelengths to model the Mg II line profiles and infer the properties of the CSM shells. We find that the CSM shell of SN2020xga resides at $\sim 1.3 \times 10^{16} \rm cm$ moving with a maximum velocity of $4275~\rm km~s^{-1}$, and the shell of SN2022xgc is located at $\sim 0.8 \times 10^{16} \rm cm$ reaching up to $4400~\rm km~s^{-1}$. These shells were expelled $\sim 11$ and $\sim 5$ months before explosion for SN2020xga and SN2022xgc respectively, possibly as a result of Luminous Blue Variable-like eruptions or pulsational pair instability (PPI) mass loss. We also analyze optical photometric data and model the light curves considering powering from the magnetar spin-down mechanism. The results support very energetic magnetars, approaching the mass-shedding limit, powering these SNe with ejecta masses of $\sim 7-9 \rm~M_\odot$. The ejecta masses inferred from the magnetar modeling are not consistent with the PPI scenario pointing towards stars $> 50~\rm M_\odot$ He-core, hence alternative scenarios such as fallback accretion are discussed., Comment: 25 pages text, 8 pages appendix, 21 figures. Submitted to A&A

Manual parallelization of code remains a significant challenge due to the complexities of modern software systems and the widespread adoption of multi-core architectures. This paper introduces OMPar, an AI-driven tool designed to automate the parallelization of C/C++ code using OpenMP pragmas. OMPar integrates Large Language Models (LLMs) through two key components: OMPify, which assesses loop parallelization potential, and MonoCoder-OMP, a new fine-tuned model which generates precise OpenMP pragmas. The evaluation of OMPar follows the same rigorous process applied to traditional tools like source-to-source AutoPar and ICPC compilers: (1) ensuring the generated code compiles and runs correctly in serial form, (2) assessing performance with the gradual addition of threads and corresponding physical cores, and (3) verifying and validating the correctness of the code's output. Benchmarks from HeCBench and ParEval are used to evaluate accuracy and performance. Experimental results demonstrate that OMPar significantly outperforms traditional methods, achieving higher accuracy in identifying parallelizable loops and generating efficient pragmas. Beyond accuracy, OMPar offers advantages such as the ability to work on partial or incomplete codebases and the capacity to continuously learn from new code patterns, enhancing its parallelization capabilities over time. These results underscore the potential of LLMs in revolutionizing automatic parallelization techniques, paving the way for more efficient and scalable parallel computing systems.

Crafting a single, versatile physics-based controller that can breathe life into interactive characters across a wide spectrum of scenarios represents an exciting frontier in character animation. An ideal controller should support diverse control modalities, such as sparse target keyframes, text instructions, and scene information. While previous works have proposed physically simulated, scene-aware control models, these systems have predominantly focused on developing controllers that each specializes in a narrow set of tasks and control modalities. This work presents MaskedMimic, a novel approach that formulates physics-based character control as a general motion inpainting problem. Our key insight is to train a single unified model to synthesize motions from partial (masked) motion descriptions, such as masked keyframes, objects, text descriptions, or any combination thereof. This is achieved by leveraging motion tracking data and designing a scalable training method that can effectively utilize diverse motion descriptions to produce coherent animations. Through this process, our approach learns a physics-based controller that provides an intuitive control interface without requiring tedious reward engineering for all behaviors of interest. The resulting controller supports a wide range of control modalities and enables seamless transitions between disparate tasks. By unifying character control through motion inpainting, MaskedMimic creates versatile virtual characters. These characters can dynamically adapt to complex scenes and compose diverse motions on demand, enabling more interactive and immersive experiences., Comment: ACM Transactions on Graphics (Proc. SIGGRAPH Asia 2024) Project page: https://research.nvidia.com/labs/par/maskedmimic/

Growing observational evidence suggests that enhanced mass loss from the progenitors of core-collapse supernovae (SNe) is common during $\sim1$ yr preceding the explosion, creating an optically thick circum-stellar medium (CSM) shell at $\sim10^{14.5}$ cm radii. We show that if such mass loss is indeed common, then the breakout of the SN shock through the dense CSM shell produces a neutrino flux that may account for a significant fraction of the observed $\gtrsim10$ TeV neutrino background. The neutrinos are created within a few days from the explosion, during and shortly after the shock breakout, which produces also large UV (and later X-ray) luminosity. The compact size and large UV luminosity imply a pair production optical depth of $\sim10^4$ for $>100$ GeV photons, naturally accounting for the lack of a high-energy gamma-ray background accompanying the neutrino background. SNe producing $>1$ neutrino event in a 1 km$^2$ detector are expected at a rate of $\lesssim0.1$/yr. A quantitative theory describing the evolution of the electromagnetic spectrum during a breakout, as the radiation-mediated shock is transformed into a collisionless one, is required to enable (i) using data from upcoming surveys that will systematically detect large numbers of young, $<1$ d old SNe, to determine the pre-explosion mass loss history of the SN progenitor population, and (ii) a quantitative determination of the neutrino luminosity and spectrum.

SN 2023tsz is a Type Ibn supernova (SNe Ibn) discovered in an extremely low-mass host. SNe Ibn are an uncommon subtype of stripped-envelope core-collapse SNe. They are characterised by narrow helium emission lines in their spectra and are believed to originate from the collapse of massive Wolf-Rayet (WR) stars, though their progenitor systems still remain poorly understood. In terms of energetics and spectrophotometric evolution, SN 2023tsz is largely a typical example of the class, although line profile asymmetries in the nebular phase are seen, which may indicate the presence of dust formation or unshocked circumstellar material. Intriguingly, SN 2023tsz is located in an extraordinarily low-mass host galaxy that is in the 2nd percentile for SESN host masses and star formation rates (SFR). The host has a radius of 1.0 kpc, a $g$-band absolute magnitude of $-12.73$, and an estimated metallicity of $\log(Z_{*}/Z_{\odot}$) = $-1.56$. The SFR and metallicity of the host galaxy raise questions about the progenitor of SN 2023tsz. The low SFR suggests that a star with sufficient mass to evolve into a WR would be uncommon in this galaxy. Further, the very low-metallicity is a challenge for single stellar evolution to enable H and He stripping of the progenitor and produce a SN Ibn explosion. The host galaxy of SN 2023tsz adds another piece to the ongoing puzzle of SNe Ibn progenitors, and demonstrates that they can occur in hosts too faint to be observed in contemporary sky surveys at a more typical SN Ibn redshift., Comment: 14 pages, 10 figures, submitted to MNRAS

A family of subsets $\mathcal{F} \subseteq \mathcal{P}(\{1, 2, \ldots, n\})$ has the disparate union property if any two disjoint subfamilies $\mathcal{F}_1, \mathcal{F}_2 \subseteq \mathcal{F}$ have distinct unions $\bigcup \mathcal{F}_1 \neq \bigcup \mathcal{F}_2$; what is the maximal size of a family with the disparate union property? Is there a simple and efficiently computable characterization of size-maximal families? This paper highlights a class of partially-ordered semirings -- difference ordered semirings with a multiplicatively absorbing element -- and shows it is common and easily constructed. We prove that a suitably modified definition of linear independence for semimodules over such semirings enjoys the same maximality property as for vector spaces, and can furthermore be efficiently detected by the bideterminant. These properties allow us to extend dimension argument in extremal combinatorics and provide simple and direct solutions to the puzzles above.

We report the first measurement of the azimuthal anisotropy of J$/\psi$ at forward rapidity ($1.2<|\eta|<2.2$) in Au$+$Au collisions at $\sqrt{s_{_{NN}}}=200$ GeV at the Relativistic Heavy Ion Collider. The data were collected by the PHENIX experiment in 2014 and 2016 with integrated luminosity of 14.5~nb$^{-1}$. The second Fourier coefficient ($v_2$) of the azimuthal distribution of $J/\psi$ is determined as a function of the transverse momentum ($p_T$) using the event-plane method. The measurements were performed for several selections of collision centrality: 0\%--50\%, 10\%--60\%, and 10\%-40\%. We find that in all cases the values of $v_2(p_T)$, which quantify the elliptic flow of J$/\psi$, are consistent with zero. The results are consistent with measurements at midrapidity, indicating no significant elliptic flow of the J$/\psi$ within the quark-gluon-plasma medium at collision energies of $\sqrt{s_{_{NN}}}=200$ GeV., Comment: 369 authors from 72 institutions, 12 pages, 7 figures, 5 tables. v1 is version submitted to Physical Review C. HEPdata tables for the points plotted in figures for this and previous PHENIX publications are (or will be) publicly available at http://www.phenix.bnl.gov/papers.html

We present the first forward-rapidity measurements of elliptic anisotropy of open-heavy-flavor muons at the BNL Relativistic Heavy Ion Collider. The measurements are based on data samples of Au$+$Au collisions at $\sqrt{s_{_{NN}}}=200$ GeV collected by the PHENIX experiment in 2014 and 2016 with integrated luminosity of 14.5~nb$^{-1}$. The measurements are performed in the pseudorapidity range $1.2<|\eta|<2$ and cover transverse momenta $1

Generative AI (GenAI) has been shown to enhance individual productivity in a guided setting. While it is also likely to transform processes in a collaborative work setting, it is unclear what trajectory this transformation will follow. Collaborative environment is characterized by a blend of origination tasks that involve building something from scratch and iteration tasks that involve refining on others' work. Whether GenAI affects these two aspects of collaborative work and to what extent is an open empirical question. We study this question within the open-source development landscape, a prime example of collaborative innovation, where contributions are voluntary and unguided. Specifically, we focus on the launch of GitHub Copilot in October 2021 and leverage a natural experiment in which GitHub Copilot (a programming-focused LLM) selectively rolled out support for Python, but not for R. We observe a significant jump in overall contributions, suggesting that GenAI effectively augments collaborative innovation in an unguided setting. Interestingly, Copilot's launch increased maintenance-related contributions, which are mostly iterative tasks involving building on others' work, significantly more than code-development contributions, which are mostly origination tasks involving standalone contributions. This disparity was exacerbated in active projects with extensive coding activity, raising concerns that, as GenAI models improve to accommodate richer context, the gap between origination and iterative solutions may widen. We discuss practical and policy implications to incentivize high-value innovative solutions., Comment: JEL Classification: O31, C88, J24, O35, L86

We consider the overfitting behavior of minimum norm interpolating solutions of Gaussian kernel ridge regression (i.e. kernel ridgeless regression), when the bandwidth or input dimension varies with the sample size. For fixed dimensions, we show that even with varying or tuned bandwidth, the ridgeless solution is never consistent and, at least with large enough noise, always worse than the null predictor. For increasing dimension, we give a generic characterization of the overfitting behavior for any scaling of the dimension with sample size. We use this to provide the first example of benign overfitting using the Gaussian kernel with sub-polynomial scaling dimension. All our results are under the Gaussian universality ansatz and the (non-rigorous) risk predictions in terms of the kernel eigenstructure.

The $J/\psi$ and $\psi(2S)$ charmonium states, composed of $c\bar{c}$ quark pairs and known since the 1970s, are widely believed to serve as ideal probes to test quantum chromodynamics in high-energy hadronic interactions. However, there is not yet a complete understanding of the charmonium-production mechanism. Recent measurements of $J/\psi$ production as a function of event charged-particle multiplicity at the collision energies of both the Large Hadron Collider (LHC) and the Relativistic Heavy Ion Collider (RHIC) show enhanced $J/\psi$ production yields with increasing multiplicity. One potential explanation for this type of dependence is multiparton interactions (MPI). We carry out the first measurements of self-normalized $J/\psi$ yields and the $\psi(2S)$ to $J/\psi$ ratio at both forward and backward rapidities as a function of self-normalized charged-particle multiplicity in $p$$+$$p$ collisions at $\sqrt{s}=200$ GeV. In addition, detailed {\sc pythia} studies tuned to RHIC energies were performed to investigate the MPI impacts. We find that the PHENIX data at RHIC are consistent with recent LHC measurements and can only be described by {\sc pythia} calculations that include MPI effects. The forward and backward $\psi(2S)$ to $J/\psi$ ratio, which serves as a unique and powerful approach to study final-state effects on charmonium production, is found to be less dependent on the charged-particle multiplicity., Comment: 301 authors from 69 institutions, 8 pages, 3 figures. v1 is version submitted to Physical Review D Letters. HEPdata tables for the points plotted in figures for this and previous PHENIX publications are (or will be) publicly available at http://www.phenix.bnl.gov/papers.html

We conduct a study on the relationship between galaxy environments and their active galactic nuclei (AGN) activity at high redshifts ($2.010^{12.8}M_\odot$) overdense ($\sigma_\delta>5.0$) peak, normalized with respect to the size of the peak. These findings indicate that the prevalence of AGN activity is highly dependent on the environment in which a host galaxy resides, even at early times in the formation history of the Universe., Comment: 15 pages, 11 figures, submitted to A&A, comments welcome

Strongly interacting electrons in a topologically non trivial band may form exotic phases of matter. An especially intriguing example of which is the fractional quantum anomalous Hall phase, recently discovered in twisted transition metal dichalcogenides and in moir\'e graphene multilayers. However, it has been shown to be destabilized in certain filling factors at sub-100 mK temperatures in pentalayer graphene, in favor of a novel integer quantum anomalous Hall phase [Z. Lu et al., arXiv:2408.10203 ]. We propose that the culprit stabilizing the fractional phase at higher temperatures is its rich edge state structure. Possessing a multiplicity of chiral modes on its edge, the fractional phase has lower free energy at higher temperatures due to the excess edge modes entropy. We make distinct predictions under this scenario, including the system-size dependency of the fractional phase entropic enhancement, and how the phase boundaries change as a function of temperature.

A many body system in the vicinity of a first-order phase transition may get trapped in a local minimum of the free energy landscape. These so-called false-vacuum states may survive for exceedingly long times if the barrier for their decay is high enough. The rich phase diagram obtained in graphene multilayer devices presents a unique opportunity to explore transient superconductivity on top of a correlated false vacuum. Specifically, we consider superconductors which are terminated by an apparent first-order phase transition to a correlated phase with different symmetry. We propose that quenching across this transition leads to a non-equilibrium ephemeral superconductor, readily detectable using straightforward transport measurements. Besides enabling a simple detection scheme, the transient superconductor also generically enhances the false vacuum lifetime, potentially by orders of magnitude. In several scenarios, the complimentary effect takes place as well: superconductivity is temporarily emboldened in the false vacuum, albeit ultimately decaying. We demonstrate the applicability of these claims for two different instances of superconductivity terminated by a first order transition in rhombohedral graphene. The obtained decay timescales position this class of materials as a promising playground to unambiguously realize and measure non-equilibrium superconductivity.

The cores of stars are the cosmic furnaces where light elements are fused into heavier nuclei. The fusion of hydrogen to helium initially powers all stars. The ashes of the fusion reactions are then predicted to serve as fuel in a series of stages, eventually transforming massive stars into a structure of concentric shells. These are composed of natal hydrogen on the outside, and consecutively heavier compositions inside, predicted to be dominated by helium, carbon/oxygen, oxygen/neon/magnesium, and oxygen/silicon/sulphur. Silicon and sulphur are fused into inert iron, leading to the collapse of the core and either a supernova explosion or the direct formation of a black hole. Stripped stars, where the outer hydrogen layer has been removed and the internal He-rich layer (in Wolf-Rayet WN stars) or even the C/O layer below it (in Wolf-Rayet WC/WO stars) are exposed, provide evidence for this shell structure, and the cosmic element production mechanism it reflects. The types of supernova explosions that arise from stripped stars embedded in shells of circumstellar material (most notably Type Ibn supernovae from stars with outer He layers, and Type Icn supernovae from stars with outer C/O layers) confirm this scenario. However, direct evidence for the most interior shells, which are responsible for the production of elements heavier than oxygen, is lacking. Here, we report the discovery of the first-of-its-kind supernova arising from a star peculiarly stripped all the way to the silicon and sulphur-rich internal layer. Whereas the concentric shell structure of massive stars is not under debate, it is the first time that such a thick, massive silicon and sulphur-rich shell, expelled by the progenitor shortly before the SN explosion, has been directly revealed., Comment: 48 pages, 12 figures and 10 tables. Submitted to a high-impact journal. The reduced spectra and photometry will be made available via the journal webpage and the WISeREP archive after the acceptance of the paper

A Reflective Intelligent Surface (RIS) consists of many small reflective elements whose reflection properties can be adjusted to change the wireless propagation environment. Envisioned implementations require that each RIS element be connected to a controller, and as the number of RIS elements on a surface may be on the order of hundreds or more, the number of required electrical connectors creates a difficult wiring problem, especially at high frequencies where the physical space between the elements is limited. A potential solution to this problem was previously proposed by the authors in which "biasing transmission lines" carrying standing waves are sampled at each RIS location to produce the desired bias voltage for each RIS element. This solution has the potential to substantially reduce the complexity of the RIS control. This paper presents models for the RIS elements that account for mutual coupling and realistic varactor characteristics, as well as circuit models for sampling the transmission line to generate the RIS control signals. For the latter case, the paper investigates two techniques for conversion of the transmission line standing wave voltage to the varactor bias voltage, namely an envelope detector and a sample-and-hold circuit. The paper also develops a modal decomposition approach for generating standing waves that are able to generate beams and nulls in the resulting RIS radiation pattern that maximize either the Signal-to-Noise Ratio (SNR) or the Signal-to-Leakage-plus-Noise Ratio (SLNR). Extensive simulation results are provided for the two techniques, together with a discussion of computational complexity., Comment: 20 pages, 25 figures, 2 tables

We present Jamba-1.5, new instruction-tuned large language models based on our Jamba architecture. Jamba is a hybrid Transformer-Mamba mixture of experts architecture, providing high throughput and low memory usage across context lengths, while retaining the same or better quality as Transformer models. We release two model sizes: Jamba-1.5-Large, with 94B active parameters, and Jamba-1.5-Mini, with 12B active parameters. Both models are fine-tuned for a variety of conversational and instruction-following capabilties, and have an effective context length of 256K tokens, the largest amongst open-weight models. To support cost-effective inference, we introduce ExpertsInt8, a novel quantization technique that allows fitting Jamba-1.5-Large on a machine with 8 80GB GPUs when processing 256K-token contexts without loss of quality. When evaluated on a battery of academic and chatbot benchmarks, Jamba-1.5 models achieve excellent results while providing high throughput and outperforming other open-weight models on long-context benchmarks. The model weights for both sizes are publicly available under the Jamba Open Model License and we release ExpertsInt8 as open source., Comment: Webpage: https://www.ai21.com/jamba

Recent advances in self-supervised learning enabled novel medical AI models, known as foundation models (FMs) that offer great potential for characterizing health from diverse biomedical data. Continuous glucose monitoring (CGM) provides rich, temporal data on glycemic patterns, but its full potential for predicting broader health outcomes remains underutilized. Here, we present GluFormer, a generative foundation model on biomedical temporal data based on a transformer architecture, and trained on over 10 million CGM measurements from 10,812 non-diabetic individuals. We tokenized the CGM training data and trained GluFormer using next token prediction in a generative, autoregressive manner. We demonstrate that GluFormer generalizes effectively to 15 different external datasets, including 4936 individuals across 5 different geographical regions, 6 different CGM devices, and several metabolic disorders, including normoglycemic, prediabetic, and diabetic populations, as well as those with gestational diabetes and obesity. GluFormer produces embeddings which outperform traditional CGM analysis tools, and achieves high Pearson correlations in predicting clinical parameters such as HbA1c, liver-related parameters, blood lipids, and sleep-related indices. Notably, GluFormer can also predict onset of future health outcomes even 4 years in advance. We also show that CGM embeddings from pre-intervention periods in Randomized Clinical Trials (RCTs) outperform other methods in predicting primary and secondary outcomes. When integrating dietary data into GluFormer, we show that the enhanced model can accurately generate CGM data based only on dietary intake data, simulate outcomes of dietary interventions, and predict individual responses to specific foods. Overall, we show that GluFormer accurately predicts health outcomes which generalize across different populations metabolic conditions.

Controlling humanoids in complex physically simulated worlds is a long-standing challenge with numerous applications in gaming, simulation, and visual content creation. In our setup, given a rich and complex 3D scene, the user provides a list of instructions composed of target locations and locomotion types. To solve this task we present PlaMo, a scene-aware path planner and a robust physics-based controller. The path planner produces a sequence of motion paths, considering the various limitations the scene imposes on the motion, such as location, height, and speed. Complementing the planner, our control policy generates rich and realistic physical motion adhering to the plan. We demonstrate how the combination of both modules enables traversing complex landscapes in diverse forms while responding to real-time changes in the environment. Video: https://youtu.be/wWlqSQlRZ9M .

Text-to-image diffusion models have proven effective for solving many image editing tasks. However, the seemingly straightforward task of seamlessly relocating objects within a scene remains surprisingly challenging. Existing methods addressing this problem often struggle to function reliably in real-world scenarios due to lacking spatial reasoning. In this work, we propose a training-free method, dubbed DiffUHaul, that harnesses the spatial understanding of a localized text-to-image model, for the object dragging task. Blindly manipulating layout inputs of the localized model tends to cause low editing performance due to the intrinsic entanglement of object representation in the model. To this end, we first apply attention masking in each denoising step to make the generation more disentangled across different objects and adopt the self-attention sharing mechanism to preserve the high-level object appearance. Furthermore, we propose a new diffusion anchoring technique: in the early denoising steps, we interpolate the attention features between source and target images to smoothly fuse new layouts with the original appearance; in the later denoising steps, we pass the localized features from the source images to the interpolated images to retain fine-grained object details. To adapt DiffUHaul to real-image editing, we apply a DDPM self-attention bucketing that can better reconstruct real images with the localized model. Finally, we introduce an automated evaluation pipeline for this task and showcase the efficacy of our method. Our results are reinforced through a user preference study., Comment: Accepted to SIGGRAPH Asia 2024. Project page is available at https://omriavrahami.com/diffuhaul/

Hydrogen-rich superluminous supernovae (SLSNe II) are rare. The exact mechanism producing their extreme light curve peaks is not understood. Analysis of single events and small samples suggest that CSM interaction is the main responsible for their features. However, other mechanisms can not be discarded. Large sample analysis can provide clarification. We aim to characterize the light curves of a sample of 107 SLSNe II to provide valuable information that can be used to validate theoretical models. We analyze the gri light curves of SLSNe II obtained through ZTF. We study peak absolute magnitudes and characteristic timescales. When possible we compute g-r colors, pseudo-bolometric light curves, and estimate lower limits for their total radiated energy. We also study the luminosity distribution of our sample and estimate the percentage of them that would be observable by the LSST. Finally, we compare our sample to other H-rich SNe and to H-poor SLSNe I. SLSNe II are heterogeneous. Their median peak absolute magnitude is -20.3 mag in optical bands. Their rise can take from two weeks to over three months, and their decline from twenty days to over a year. We found no significant correlations between peak magnitude and timescales. SLSNe II tend to show fainter peaks, longer declines and redder colors than SLSNe I. We present the largest sample of SLSNe II light curves to date, comprising of 107 events. Their diversity could be explained by considering different CSM morphologies. Although, theoretical analysis is needed to explore alternative scenarios. Other luminous transients, such as Active Galactic Nuclei, Tidal Disruption Events or SNe Ia-CSM, can easily become contaminants. Thus, good multi-wavelength light curve coverage becomes paramount. LSST could miss 30 percent of the ZTF events in the its footprint in gri bands. Redder bands become important to construct complete samples., Comment: 35 pages. 27 figures. 9 tables

We study a magnitude-limited sample of 36 Broad-lined Type Ic Supernovae (SNe Ic-BL) from the Zwicky Transient Facility Bright Transient Survey (detected between March 2018 and August 2021), which is the largest systematic study of SNe Ic-BL done in literature thus far. We present the light curves (LCs) for each of the SNe, and analyze the shape of the LCs to derive empirical parameters, along with the explosion epochs for every event. The sample has an average absolute peak magnitude in the r band of $M_r^{max}$ = -18.51 $\pm$ 0.15 mag. Using spectra obtained around peak light, we compute expansion velocities from the Fe II 5169 Angstrom line for each event with high enough signal-to-noise ratio spectra, and find an average value of $v_{ph}$ = 16,100 $\pm$ 1,100 km $s^{-1}$. We also compute bolometric LCs, study the blackbody temperature and radii evolution over time, and derive the explosion properties of the SNe. The explosion properties of the sample have average values of $M_{Ni}$ = $0.37_{-0.06}^{+0.08}$ solar masses, $M_{ej}$ = $2.45_{-0.41}^{+0.47}$ solar masses, and $E_K$= $4.02_{-1.00}^{+1.37} \times 10^{51}$ erg. Thirteen events have radio observations from the Very Large Array, with 8 detections and 5 non-detections. We find that the populations that have radio detections and radio non-detections are indistinct from one another with respect to their optically-inferred explosion properties, and there are no statistically significant correlations present between the events' radio luminosities and optically-inferred explosion properties. This provides evidence that the explosion properties derived from optical data alone cannot give inferences about the radio properties of SNe Ic-BL, and likely their relativistic jet formation mechanisms., Comment: 52 pages, 34 Figures, 8 Tables; Accepted to ApJ

Hybrid hospitals offer on-site and remote hospitalization through telemedicine. These new healthcare models require novel operational policies to balance costs, efficiency, and patient well-being. Our study addresses two first-order questions: (i) how to direct patient admission and call-in based on individual characteristics and proximity and (ii) how to determine the optimal allocation of medical resources between these two hospitalization options and among different patient types. We develop a model that uses Brownian Motion to capture the patient's health evolution during remote/on-site hospitalization and during travel. Under cost-minimizing call-in policies, we find that remote hospitalization can be cost-effective for moderately distant patients, as the optimal call-in threshold is non-monotonic in the patient's travel time. Subject to scarce resources, the optimal solution structure becomes equivalent to a simultaneous, identically sized increase of remote and on-site costs under abundant resources. When limited resources must be divided among multiple patient types, the optimal thresholds shift in non-obvious ways as resource availability changes. Finally, we develop a practical and efficient policy that allows for swapping an on-site patient with a remote patient when the latter is called-in and sufficient resources are not available to treat both on-site. Contrary to the widely held view that telemedicine can mitigate rural and non-rural healthcare disparities, our research suggests that on-site care may actually be more cost-effective than remote hospitalization for patients in distant locations, due to (potentially overlooked) risks during patient travel. This finding may be of particular concern in light of the growing number of ``hospital deserts'' amid recent rural hospital closures, as these communities may in fact not be well-served through at-home care.

We present optical and near-infrared observations of two Type Ibn supernovae (SNe), SN 2018jmt and SN 2019cj. Their light curves have rise times of about 10 days, reaching an absolute peak magnitude of $M_g$(SN 2018jmt) = $-$19.07 $\pm$ 0.37 and $M_V$(SN 2019cj) = $-$18.94 $\pm$ 0.19 mag, respectively. The early-time spectra of SN 2018jmt are dominated by a blue continuum, accompanied by narrow (600$-$1000 km~s$^{-1}$) He I lines with P-Cygni profile. At later epochs, the spectra become more similar to those of the prototypical SN Ibn 2006jc. At early phases, the spectra of SN 2019cj show flash ionisation emission lines of C III, N III and He II superposed on a blue continuum. These features disappear after a few days, and then the spectra of SN 2019cj evolve similarly to those of SN 2018jmt. The spectra indicate that the two SNe exploded within a He-rich circumstellar medium (CSM) lost by the progenitors a short time before the explosion. We model the light curves of the two SNe Ibn to constrain the progenitor and the explosion parameters. The ejecta masses are consistent with either that expected for a canonical SN Ib ($\sim$ 2 M$_{\odot}$) or those from a massive WR star ($>$ $\sim$ 4 M$_{\odot}$), with the kinetic energy on the order of $10^{51}$ erg. The lower limit on the ejecta mass ($>$ $\sim$ 2 M$_{\odot}$) argues against a scenario involving a relatively low-mass progenitor (e.g., $M_{ZAMS}$ $\sim$ 10 M$_{\odot}$). We set a conservative upper limit of $\sim$0.1 M$_{\odot}$ for the $^{56}$Ni masses in both SNe. From the light curve modelling, we determine a two-zone CSM distribution, with an inner, flat CSM component, and an outer CSM with a steeper density profile. The physical properties of SN 2018jmt and SN 2019cj are consistent with those expected from the core collapse of relatively massive, stripped-envelope (SE) stars., Comment: 28 pages, 19 figures, Accepted for publication in Astronomy & Astrophysics

We propose an empirical approach centered on the spectral dynamics of weights -- the behavior of singular values and vectors during optimization -- to unify and clarify several phenomena in deep learning. We identify a consistent bias in optimization across various experiments, from small-scale ``grokking'' to large-scale tasks like image classification with ConvNets, image generation with UNets, speech recognition with LSTMs, and language modeling with Transformers. We also demonstrate that weight decay enhances this bias beyond its role as a norm regularizer, even in practical systems. Moreover, we show that these spectral dynamics distinguish memorizing networks from generalizing ones, offering a novel perspective on this longstanding conundrum. Additionally, we leverage spectral dynamics to explore the emergence of well-performing sparse subnetworks (lottery tickets) and the structure of the loss surface through linear mode connectivity. Our findings suggest that spectral dynamics provide a coherent framework to better understand the behavior of neural networks across diverse settings.