Search

Large Language Models (LLMs) for public use require continuous pre-training to remain up-to-date with the latest data. The models also need to be fine-tuned with specific instructions to maintain their ability to follow instructions accurately. Typically, LLMs are released in two versions: the Base LLM, pre-trained on diverse data, and the instruction-refined LLM, additionally trained with specific instructions for better instruction following. The question arises as to which model should undergo continuous pre-training to maintain its instruction-following abilities while also staying current with the latest data. In this study, we delve into the intricate relationship between continuous pre-training and instruction fine-tuning of the LLMs and investigate the impact of continuous pre-training on the instruction following abilities of both the base and its instruction finetuned model. Further, the instruction fine-tuning process is computationally intense and requires a substantial number of hand-annotated examples for the model to learn effectively. This study aims to find the most compute-efficient strategy to gain up-to-date knowledge and instruction-following capabilities without requiring any instruction data and fine-tuning. We empirically prove our findings on the LLaMa 3, 3.1 and Qwen 2, 2.5 family of base and instruction models, providing a comprehensive exploration of our hypotheses across varying sizes of pre-training data corpus and different LLMs settings.

Federated learning faces a critical challenge in balancing communication efficiency with rapid convergence, especially for second-order methods. While Newton-type algorithms achieve linear convergence in communication rounds, transmitting full Hessian matrices is often impractical due to quadratic complexity. We introduce Federated Learning with Enhanced Nesterov-Newton Sketch (FLeNS), a novel method that harnesses both the acceleration capabilities of Nesterov's method and the dimensionality reduction benefits of Hessian sketching. FLeNS approximates the centralized Newton's method without relying on the exact Hessian, significantly reducing communication overhead. By combining Nesterov's acceleration with adaptive Hessian sketching, FLeNS preserves crucial second-order information while preserving the rapid convergence characteristics. Our theoretical analysis, grounded in statistical learning, demonstrates that FLeNS achieves super-linear convergence rates in communication rounds - a notable advancement in federated optimization. We provide rigorous convergence guarantees and characterize tradeoffs between acceleration, sketch size, and convergence speed. Extensive empirical evaluation validates our theoretical findings, showcasing FLeNS's state-of-the-art performance with reduced communication requirements, particularly in privacy-sensitive and edge-computing scenarios. The code is available at https://github.com/sunnyinAI/FLeNS, Comment: 10 pages, 3 figures, 2 Tables

Noncentrosymmetric 2D superconductors with large spin-orbit coupling offer an opportunity to explore superconducting behaviors far beyond the Pauli limit. One such superconductor, few-layer T$_d$-MoTe$_2$, has large upper critical fields that can exceed the Pauli limit by up to 600%. However, the mechanisms governing this enhancement are still under debate, with theory pointing towards either spin-orbit parity coupling or tilted Ising spin-orbit coupling. Moreover, ferroelectricity concomitant with superconductivity has been recently observed in the bilayer, where strong changes to superconductivity can be observed throughout the ferroelectric transition pathway. Here, we report the superconducting behavior of bilayer T$_d$-MoTe$ _2$ under an in-plane magnetic field, while systematically varying magnetic field angle and out-of-plane electric field strength. We find that superconductivity in bilayer MoTe$_2$ exhibits a two-fold symmetry with an upper critical field maxima occurring along the b-axis and minima along the a-axis. The two-fold rotational symmetry remains robust throughout the entire superconducting region and ferroelectric hysteresis loop. Our experimental observations of the spin-orbit coupling strength (up to 16.4 meV) agree with the spin texture and spin splitting from first-principles calculations, indicating that tilted Ising spin-orbit coupling is the dominant underlying mechanism.

Semantic segmentation of medical images is pivotal in applications like disease diagnosis and treatment planning. While deep learning has excelled in automating this task, a major hurdle is the need for numerous annotated segmentation masks, which are resource-intensive to produce due to the required expertise and time. This scenario often leads to ultra low-data regimes, where annotated images are extremely limited, posing significant challenges for the generalization of conventional deep learning methods on test images. To address this, we introduce a generative deep learning framework, which uniquely generates high-quality paired segmentation masks and medical images, serving as auxiliary data for training robust models in data-scarce environments. Unlike traditional generative models that treat data generation and segmentation model training as separate processes, our method employs multi-level optimization for end-to-end data generation. This approach allows segmentation performance to directly influence the data generation process, ensuring that the generated data is specifically tailored to enhance the performance of the segmentation model. Our method demonstrated strong generalization performance across 9 diverse medical image segmentation tasks and on 16 datasets, in ultra-low data regimes, spanning various diseases, organs, and imaging modalities. When applied to various segmentation models, it achieved performance improvements of 10-20\% (absolute), in both same-domain and out-of-domain scenarios. Notably, it requires 8 to 20 times less training data than existing methods to achieve comparable results. This advancement significantly improves the feasibility and cost-effectiveness of applying deep learning in medical imaging, particularly in scenarios with limited data availability.

The rise of machine learning workload on smartphones has propelled GPUs into one of the most power-hungry components of modern smartphones and elevates the need for optimizing the GPU power draw by mobile apps. Optimizing the power consumption of mobile GPUs in turn requires accurate estimation of their power draw during app execution. In this paper, we observe that the prior-art, utilization-frequency based GPU models cannot capture the diverse micro-architectural usage of modern mobile GPUs.We show that these models suffer poor modeling accuracy under diverse GPU workload, and study whether performance monitoring counter (PMC)-based models recently proposed for desktop/server GPUs can be applied to accurately model mobile GPU power. Our study shows that the PMCs that come with dominating mobile GPUs used in modern smartphones are sufficient to model mobile GPU power, but exhibit multicollinearity if used altogether. We present APGPM, the mobile GPU power modeling methodology that automatically selects an optimal set of PMCs that maximizes the GPU power model accuracy. Evaluation on two representative mobile GPUs shows that APGPM-generated GPU power models reduce the MAPE modeling error of prior-art by 1.95x to 2.66x (i.e., by 11.3% to 15.4%) while using only 4.66% to 20.41% of the total number of available PMCs.

We calculate the deformed and non-deformed cohomological Hall algebra (CoHA) of the preprojective algebra for the case of cyclic quivers by studying the Kontsevich-Soibelman CoHA and using tools from cohomological Donaldson-Thomas theory. We show that for the cyclic quiver of length $K$, this algebra is the universal enveloping algebra of the positive half of a certain extension of matrix differential operators on $\mathbb{C}^{*}$, while its deformation gives a positive half of an explicit integral form of Guay's Affine Yangian $\ddot{\mathcal{Y}}_{\hbar_1,\hbar_2}(\mathfrak{gl}(K))$. By the main theorem of Botta-Davison (2023) and Schiffmann-Vasserot (2023), this also determines the positive half of the Maulik-Okounkov Yangian for the case of cyclic quivers. Furthermore, we provide evidence for the strong rationality conjecture, calculate the spherical subalgebra of the non-deformed CoHA for any quiver without loops, recover results about the CoHA of compactly supported semistable sheaves on the minimal resolution of the Kleinian singularity $\mathbb{C}^2/\mathbb{Z}_{K}$ and identify a commutative algebra inside the additive shuffle algebra associated to the cyclic quiver. We end by conjecturally relating the obtained integral form with the algebra defined by Gaiotto-Rap\v{c}\'ak-Zhou, in the context of twisted M-theory., Comment: 52 pages, comments welcome

In Hindustani classical music, the tabla plays an important role as a rhythmic backbone and accompaniment. In applications like computer-based music analysis, learning singing, and learning musical instruments, tabla stroke transcription, $t\bar{a}la$ identification, and generation are crucial. This paper proposes a comprehensive system aimed at addressing these challenges. For tabla stroke transcription, we propose a novel approach based on model-agnostic meta-learning (MAML) that facilitates the accurate identification of tabla strokes using minimal data. Leveraging these transcriptions, the system introduces two novel $t\bar{a}la$ identification methods based on the sequence analysis of tabla strokes. \par Furthermore, the paper proposes a framework for $t\bar{a}la$ generation to bridge traditional and modern learning methods. This framework utilizes finite state transducers (FST) and linear time-invariant (LTI) filters to generate $t\bar{a}las$ with real-time tempo control through user interaction, enhancing practice sessions and musical education. Experimental evaluations on tabla solo and concert datasets demonstrate the system's exceptional performance on real-world data and its ability to outperform existing methods. Additionally, the proposed $t\bar{a}la$ identification methods surpass state-of-the-art techniques. The contributions of this paper include a combined approach to tabla stroke transcription, innovative $t\bar{a}la$ identification techniques, and a robust framework for $t\bar{a}la$ generation that handles the rhythmic complexities of Hindustani music.

For a metric space $(X,d)$, Beer, Naimpally, and Rodriguez-Lopez in ([17]) proposed a unified approach to explore set convergences via uniform convergence of distance functionals on members of an arbitrary family $\mathcal{S}$ of subsets of $X$. The associated topology on the collection $CL(X)$ of all nonempty closed subsets of $(X,d)$ is denoted by $\tau_{\mathcal{S},d}$. As special cases, this unified approach includes classical Wijsman, Attouch-Wets, and Hausdorff distance topologies. In this article, we investigate various topological characteristics of the hyperspace $(CL(X), \tau_{\mathcal{S},d})$ when $\mathcal{S}$ is a bornology on $(X,d)$. In order to do this, a new class of bornologies and a new metric topology on $CL(X)$ have been introduced and studied., Comment: 25 pages

In transition metal dichalcogenide semiconductor monolayers the spin dynamics of electrons is controlled by the original spin-valley locking effect resulting from the interplay between spin-orbit interaction and inversion asymmetry. As a consequence, for electrons occupying bottom conduction bands, a carrier spin flip occurs only if there is a simultaneous change of valley. However, very little is known about the intra-valley spin relaxation processes. In this work we have performed stationary and time-resolved photoluminescence measurements in high quality WSe$_2$ monolayers. Our experiments highlight an efficient relaxation from bright to dark excitons, due to a fast intra-valley electron transfer from the top to the bottom conduction band with opposite spins. A combination of experiments and theoretical analysis allows us to infer a spin relaxation time of about $\tau_s\sim10~$ps, driven by the interplay between $\Gamma$-valley chiral phonons and spin-orbit mixing., Comment: 10 pages, 4 figures

We analyze the possibility of symmetry-lowering induced by pseudo-Jahn--Teller interactions in six previously studied azaphenalenes that are known to have their first excited singlet state (S$_1$) lower in energy than the triplet state (T$_1$). The primary aim of this study is to explore whether Hund's rule violation is observed in these molecules when their structures are distorted from $C_{\rm 2v}$ or $D_{\rm 3h}$ point group symmetries by vibronic coupling. Along two interatomic distances connecting these point groups to their subgroups $C_{\rm s}$ or $C_{\rm 3h}$, we relaxed the other internal degrees of freedom and calculated two-dimensional potential energy subsurfaces. The many-body perturbation theory (MP2) suggests that the high-symmetry structures are the energy minima for all six systems. However, single-point energy calculations using the coupled-cluster method (CCSD(T)) indicate symmetry lowering in four cases. The singlet-triplet energy gap plotted on the potential energy surface also shows variations when deviating from high-symmetry structures. A full geometry optimization at the CCSD(T) level with the cc-pVTZ basis set reveals that the $D_{\rm 3h}$ structure of cyclazine (1AP) is a saddle point, connecting two equivalent minima of $C_{\rm 3h}$ symmetry undergoing rapid automerization. The combined effects of symmetry lowering and high-level corrections result in a nearly zero singlet-triplet gap for the $C_{\rm 3h}$ structure of cyclazine. Azaphenalenes containing nitrogen atoms at electron-deficient sites -- 2AP, 3AP, and 4AP -- exhibit more pronounced in-plane structural distortion; the effect is captured by the long-range exchange-interaction corrected DFT method, $\omega$B97XD. Excited state calculations of these systems indicate that in their low-symmetry energy minima, T$_1$ is indeed lower in energy than S$_1$, upholding the validity of Hund's rule., Comment: second version

Neural image representations have recently emerged as a promising technique for storing, streaming, and rendering visual data. Coupled with learning-based workflows, these novel representations have demonstrated remarkable visual fidelity and memory efficiency. However, existing neural image representations often rely on explicit uniform data structures without content adaptivity or computation-intensive implicit models, limiting their adoption in real-time graphics applications. Inspired by recent advances in radiance field rendering, we propose Image-GS, a content-adaptive image representation. Using anisotropic 2D Gaussians as the basis, Image-GS shows high memory efficiency, supports fast random access, and offers a natural level of detail stack. Leveraging a tailored differentiable renderer, Image-GS fits a target image by adaptively allocating and progressively optimizing a set of 2D Gaussians. The generalizable efficiency and fidelity of Image-GS are validated against several recent neural image representations and industry-standard texture compressors on a diverse set of images. Notably, its memory and computation requirements solely depend on and linearly scale with the number of 2D Gaussians, providing flexible controls over the trade-off between visual fidelity and run-time efficiency. We hope this research offers insights for developing new applications that require adaptive quality and resource control, such as machine perception, asset streaming, and content generation.

Integrative modeling enables structure determination for large macromolecular assemblies by combining data from multiple sources of experiment data with theoretical and computational predictions. Recent advancements in AI-based structure prediction and electron cryo-microscopy have sparked renewed enthusiasm for integrative modeling; structures from AI-based methods can be integrated with in situ maps to characterize large assemblies. This approach previously allowed us and others to determine the architectures of diverse macromolecular assemblies, such as nuclear pore complexes, chromatin remodelers, and cell-cell junctions. Experimental data spanning several scales was used in these studies, ranging from high-resolution data, such as X-ray crystallography and Alphafold structures, to low-resolution data, such as cryo-electron tomography maps and data from co-immunoprecipitation experiments. Two recurrent modeling challenges emerged across a range of studies. First, modeling disordered regions, which constituted a significant portion of these assemblies, necessitated the development of new methods. Second, methods needed to be developed to utilize the information from cryo-electron tomography, a timely challenge as structural biology is increasingly moving towards in situ characterization. Here, we recapitulate recent developments in the modeling of disordered proteins and the analysis of cryo-electron tomography data and highlight opportunities for method development in the context of integrative modeling.

We introduce a framework for the use of large language models (LLMs) in Building Understandable Messaging for Policy and Evidence Review (BUMPER). LLMs are proving capable of providing interfaces for understanding and synthesizing large databases of diverse media. This presents an exciting opportunity to supercharge the translation of scientific evidence into policy and action, thereby improving livelihoods around the world. However, these models also pose challenges related to access, trust-worthiness, and accountability. The BUMPER framework is built atop a scientific knowledge base (e.g., documentation, code, survey data) by the same scientists (e.g., individual contributor, lab, consortium). We focus on a solution that builds trustworthiness through transparency, scope-limiting, explicit-checks, and uncertainty measures. LLMs are rapidly being adopted and consequences are poorly understood. The framework addresses open questions regarding the reliability of LLMs and their use in high-stakes applications. We provide a worked example in health policy for a model designed to inform measles control programs. We argue that this framework can facilitate accessibility of and confidence in scientific evidence for policymakers, drive a focus on policy-relevance and translatability for researchers, and ultimately increase and accelerate the impact of scientific knowledge used for policy decisions., Comment: 21 pages, 6 figures

Let $ B_{\ell}(n)$ denote the number of $\ell$-regular bipartitions of $n.$ In this article, we prove that $ B_{\ell}(n)$ is always almost divisible by $p_i^j$ if $p_i^{2a_i}\geq \ell,$ where $j$ is a fixed positive integer and $\ell=p_1^{a_1}p_2^{a_2}\ldots p_m^{a_m},$ where $p_i$ are prime numbers $\geq 5.$ Further, we obtain an infinities families of congruences for $B_3(n)$ and $B_5(n)$ by using Hecke eigen form theory and a result of Newman \cite{Newmann1959}. Furthermore, by applying Radu and Seller's approach, we obtain an algorithm from which we get several congruences for $B_{p}(n)$, where $p$ is a prime number.

The pentablock, denoted as $\cP,$ is defined as follows: $$\cP= \left\{ (a_{21}, {\rm tr}(A), {\rm det}(A)) : A = [a_{ij}]_{2 \times 2} \text{ with } \|A\|<1 \right\}.$$ It originated from the work of Agler--Lykova--Young in connection with a particular case of the $\mu$-synthesis problem. It is a non-convex, polynomially convex, $\mathbb{C}$-convex, star-like about the origin, and inhomogeneous domain. This paper deals with operator theory on the pentablock. We study pentablock unitaries and isometries, providing an algebraic characterization of pentablock isometries. En route, we provide the Wold-type decomposition for pentablock isometries, which consists of three parts: the unitary part, the pure part, and a new component. We define this novel component as the quasi-pentablock unitary and provide a functional model for it. Additionally, a model for a class of pure pentablock isometries has been found, along with some examples. Furthermore, a representation resembling the Beurling-Lax-Halmos paradigm has been presented for the invariant subspaces of pentablock pure isometries.

The optical emission spectra of semiconducting transition-metal dichalcogenide monolayers highlight fascinating recombination processes of charged excitons (trions). When charge tunable WSe2 monolayers are moderately doped with electrons, a strong luminescence peak emerges just below the well-understood spectral lines associated with the recombination of negatively charged bright and dark trions. Despite previous investigations, its origin remains elusive. Here, we demonstrate that this luminescence peak is the result of electron-electron assisted recombination that brightens the dark trion emission. Supporting evidence for this second-order recombination process comes from identifying the equivalent brightened emission of positively charged dark trions when the monolayer is electrostatically doped with holes. Remarkably, the discovered hole-hole assisted luminescence peak emerges in the near infrared, about 500 meV below the well-studied spectral region of excitons and trions. In addition to identifying new recombination channels of these excitonic complexes, our findings accurately determine the spin-split energies of the conduction and valence bands. Both of which play crucial roles in understanding the optical properties of WSe2 based homo- and hetero-structures., Comment: 13 pages, 4 figures

Dynamic feedback linearization-based methods allow us to design control algorithms for a fairly large class of nonlinear systems in continuous time. However, this feature does not extend to their sampled counterparts, i.e., for a given dynamically feedback linearizable continuous time system, its numerical discretization may fail to be so. In this article, we present a way to construct discretization schemes (accurate up to first order) that result in schemes that are feedback linearizable. This result is an extension of our previous work, where we had considered only static feedback linearizable systems. The result presented here applies to a fairly general class of nonlinear systems, in particular, our analysis applies to both endogenous and exogenous types of feedback. While the results in this article are presented on a control affine form of nonlinear systems, they can be readily modified to general nonlinear systems.

We present machine learning models based on kernel-ridge regression for predicting X-ray photoelectron spectra of organic molecules originating from the $K$-shell ionization energies of carbon (C), nitrogen (N), oxygen (O), and fluorine (F) atoms. We constructed the training dataset through high-throughput calculations of $K$-shell core-electron binding energies (CEBEs) for 12,880 small organic molecules in the bigQM7$\omega$ dataset, employing the $\Delta$-SCF formalism coupled with meta-GGA-DFT and a variationally converged basis set. The models are cost-effective, as they require the atomic coordinates of a molecule generated using universal force fields while estimating the target-level CEBEs corresponding to DFT-level equilibrium geometry. We explore transfer learning by utilizing the atomic environment feature vectors learned using a graph neural network framework in kernel-ridge regression. Additionally, we enhance accuracy within the $\Delta$-machine learning framework by leveraging inexpensive baseline spectra derived from Kohn--Sham eigenvalues. When applied to 208 combinatorially substituted uracil molecules larger than those in the training set, our analyses suggest that the models may not provide quantitatively accurate predictions of CEBEs but offer a strong linear correlation relevant for virtual high-throughput screening. We present the dataset and models as the Python module, ${\tt cebeconf}$, to facilitate further explorations., Comment: Major Revision, New Figures and Tables added in the SI, Figures 1 and 4 revised

While significant work has been done in the field of NLP on vertical thinking, which involves primarily logical thinking, little work has been done towards lateral thinking, which involves looking at problems from an unconventional perspective and defying existing conceptions and notions. Towards this direction, SemEval 2024 introduces the task of BRAINTEASER, which involves two types of questions -- Sentence Puzzles and Word Puzzles that defy conventional common-sense reasoning and constraints. In this paper, we tackle both types of questions using few-shot prompting on GPT-3.5 and gain insights regarding the difference in the nature of the two types. Our prompting strategy placed us 26th on the leaderboard for the Sentence Puzzle and 15th on the Word Puzzle task.

This paper develops mathematical tools to estimate seasonal changes in measles transmission rates and corresponding variation in population susceptibility. The tools are designed to leverage times series of cases in the absence of demographic data. In particular, we focus on publicly available suspected case reports from the World Health Organization (WHO), which routinely publishes country-level, monthly aggregated time series. With that as input, we show that measles epidemiologies can be characterized efficiently at global-scale, and we use our estimates to recommend context-specific, future supplementary immunization times. Throughout the paper, comparisons with more data-informed models illustrate that the approach captures the essential dynamics, and broadly speaking, the tools we describe represent a scalable intermediate between conventional empirical approaches and more intricate disease models.

This paper examines the equivalence between various set convergences, as studied in [7, 13, 22], induced by an arbitrary bornology $\mathcal{S}$ on a metric space $(X,d)$. Specifically, it focuses on the upper parts of the following set convergences: convergence deduced through uniform convergence of distance functionals on $\mathcal{S}$ ($\tau_{\mathcal{S},d}$-convergence); convergence with respect to gap functionals determined by $\mathcal{S}$ ($G_{\mathcal{S},d}$-convergence); and bornological convergence ($\mathcal{S}$-convergence). In particular, we give necessary and sufficient conditions on the structure of the bornology $\mathcal{S}$ for the coincidence of $\tau_{\mathcal{S},d}^+$-convergence with $\mathsf{G}_{\mathcal{S},d}^+$-convergence, as well as $\tau_{\mathcal{S},d}^+$-convergence with $\mathcal{S}^+$-convergence. A characterization for the equivalence of $\tau_{\mathcal{S},d}^+$-convergence and $\mathcal{S}^+$-convergence, in terms of certain convergence of nets, has also been given earlier by Beer, Naimpally, and Rodriguez-Lopez in [13]. To facilitate our study, we first devise new characterizations for $\tau_{\mathcal{S},d}^+$-convergence and $\mathcal{S}^+$-convergence, which we call their miss-type characterizations.

We present the largest optical photometry compilation of Gamma-Ray Bursts (GRBs) with redshifts ($z$). We include 64813 observations of 535 events (including upper limits) from 28 February 1997 up to 18 August 2023. We also present a user-friendly web tool \textit{grbLC} which allows users the visualization of photometry, coordinates, redshift, host galaxy extinction, and spectral indices for each event in our database. Furthermore, we have added a Gamma Ray Coordinate Network (GCN) scraper that can be used to collect data by gathering magnitudes from the GCNs. The web tool also includes a package for uniformly investigating colour evolution. We compute the optical spectral indices for 138 GRBs for which we have at least 4 filters at the same epoch in our sample and craft a procedure to distinguish between GRBs with and without colour evolution. By providing a uniform format and repository for the optical catalogue, this web-based archive is the first step towards unifying several community efforts to gather the photometric information for all GRBs with known redshifts. This catalogue will enable population studies by providing light curves (LCs) with better coverage since we have gathered data from different ground-based locations. Consequently, these LCs can be used to train future LC reconstructions for an extended inference of the redshift. The data gathering also allows us to fill some of the orbital gaps from Swift in crucial points of the LCs, e.g., at the end of the plateau emission or where a jet break is identified., Comment: 20 pages, 16 figures, 2 tables. Submitted to MNRAS, this version matches the third revision. The Online Materials and data will be available after the publication

The objective of our research is to investigate the electrocatalytic properties of novel metal-free quantum dots (QDs) composed of the recently discovered 2D material penta-CN2, with the aim of replacing costly and scarce catalysts such as Pt and Pd. Employing a first-principles density functional theory (DFT) based approach, the geometries of the three penta-CN2 quantum dots (QDs) of increasing sizes, 3 x 3, 3 x 4, 4 x 4 are optimized. Through comprehensive analysis, our research extensively explored the structural stability of penta-CN2 QDs, delved into their electronic properties, and assessed their catalytic performance concerning the Hydrogen Evolution Reaction (HER). Notably, the H-adsorbed penta-CN2 QDs exhibit a significant reduction in the HOMO-LUMO gap (E_{g}) ranging from 35% to 49% compared to the pristine QD. Next, we investigated their catalytic performance relevant to HER, using well-known descriptors: (i) adsorption energy, (ii) over-potential, (iii) Gibbs free energy and (iv) exchange current density along with the volcano curve. As far as size dependence of the catalytic performance is concerned, the value of Delta G^{(av)} is minimum for 3 x 3 penta-CN2 QD, with those of 3 x 4 and 4 x 4 QDs being slightly larger. Our calculations predict a high value of exchange current density 2.24 x 10^{-3} A-cm^{-2} for one of the sites (N11 for 3 x 3 QD), which we believe will lead to significantly enhanced HER properties. Our research outcomes hold great promise in advancing the discovery of abundant, non-toxic, and cost-effective catalysts for HER, playing a vital role in facilitating large-scale hydrogen production.

Gaze is an essential prompt for analyzing human behavior and attention. Recently, there has been an increasing interest in determining gaze direction from facial videos. However, video gaze estimation faces significant challenges, such as understanding the dynamic evolution of gaze in video sequences, dealing with static backgrounds, and adapting to variations in illumination. To address these challenges, we propose a simple and novel deep learning model designed to estimate gaze from videos, incorporating a specialized attention module. Our method employs a spatial attention mechanism that tracks spatial dynamics within videos. This technique enables accurate gaze direction prediction through a temporal sequence model, adeptly transforming spatial observations into temporal insights, thereby significantly improving gaze estimation accuracy. Additionally, our approach integrates Gaussian processes to include individual-specific traits, facilitating the personalization of our model with just a few labeled samples. Experimental results confirm the efficacy of the proposed approach, demonstrating its success in both within-dataset and cross-dataset settings. Specifically, our proposed approach achieves state-of-the-art performance on the Gaze360 dataset, improving by $2.5^\circ$ without personalization. Further, by personalizing the model with just three samples, we achieved an additional improvement of $0.8^\circ$. The code and pre-trained models are available at \url{https://github.com/jswati31/stage}., Comment: Accepted at CVPR 2024 Gaze workshop

Conversation is the most natural form of human communication, where each utterance can range over a variety of possible emotions. While significant work has been done towards the detection of emotions in text, relatively little work has been done towards finding the cause of the said emotions, especially in multimodal settings. SemEval 2024 introduces the task of Multimodal Emotion Cause Analysis in Conversations, which aims to extract emotions reflected in individual utterances in a conversation involving multiple modalities (textual, audio, and visual modalities) along with the corresponding utterances that were the cause for the emotion. In this paper, we propose models that tackle this task as an utterance labeling and a sequence labeling problem and perform a comparative study of these models, involving baselines using different encoders, using BiLSTM for adding contextual information of the conversation, and finally adding a CRF layer to try to model the inter-dependencies between adjacent utterances more effectively. In the official leaderboard for the task, our architecture was ranked 8th, achieving an F1-score of 0.1759 on the leaderboard.

For Music Information Retrieval downstream tasks, the most common audio representation is time-frequency-based, such as Mel spectrograms. In order to identify musical genres, this study explores the possibilities of a new form of audio representation one of the most usual MIR downstream tasks. Therefore, to discretely encoding music using deep vector quantization; a novel audio representation was created for the innovative generative music model i.e. Jukebox. The effectiveness of Jukebox's audio representation is compared to Mel spectrograms using a dataset that is almost equivalent to State-of-the-Art (SOTA) and an almost same transformer design. The results of this study imply that, at least when the transformers are pretrained using a very modest dataset of 20k tracks, Jukebox's audio representation is not superior to Mel spectrograms. This could be explained by the fact that Jukebox's audio representation does not sufficiently take into account the peculiarities of human hearing perception. On the other hand, Mel spectrograms are specifically created with the human auditory sense in mind.

Segment anything models (SAMs) are gaining attention for their zero-shot generalization capability in segmenting objects of unseen classes and in unseen domains when properly prompted. Interactivity is a key strength of SAMs, allowing users to iteratively provide prompts that specify objects of interest to refine outputs. However, to realize the interactive use of SAMs for 3D medical imaging tasks, rapid inference times are necessary. High memory requirements and long processing delays remain constraints that hinder the adoption of SAMs for this purpose. Specifically, while 2D SAMs applied to 3D volumes contend with repetitive computation to process all slices independently, 3D SAMs suffer from an exponential increase in model parameters and FLOPS. To address these challenges, we present FastSAM3D which accelerates SAM inference to 8 milliseconds per 128*128*128 3D volumetric image on an NVIDIA A100 GPU. This speedup is accomplished through 1) a novel layer-wise progressive distillation scheme that enables knowledge transfer from a complex 12-layer ViT-B to a lightweight 6-layer ViT-Tiny variant encoder without training from scratch; and 2) a novel 3D sparse flash attention to replace vanilla attention operators, substantially reducing memory needs and improving parallelization. Experiments on three diverse datasets reveal that FastSAM3D achieves a remarkable speedup of 527.38x compared to 2D SAMs and 8.75x compared to 3D SAMs on the same volumes without significant performance decline. Thus, FastSAM3D opens the door for low-cost truly interactive SAM-based 3D medical imaging segmentation with commonly used GPU hardware. Code is available at https://github.com/arcadelab/FastSAM3D.

LLMOps incur significant costs due to hardware requirements, hindering their widespread accessibility. Additionally, a lack of transparency in model training methods and data contributes to the majority of models being non-reproducible. To tackle these challenges, the LLM Efficiency Challenge was introduced at NeurIPS Workshop, aiming to adapt foundation models on a diverse set of tasks via fine-tuning on a single GPU (RTX 4090 or A100 with 40GB) within a 24-hour timeframe. In this system description paper, we introduce Birbal, our Mistral-7B based winning model, fine-tuned on a single RTX 4090 for 16 hours. Birbal's success lies in curating high-quality instructions covering diverse tasks, resulting in a 35% performance improvement over second-best Qwen-14B based submission.