Search

Letters of recommendation from school counselors are required to apply to many selective colleges and universities. Still, relatively little is known about how this non-standardized component may affect equity in admissions. We use cutting-edge natural language processing techniques to algorithmically analyze a national dataset of over 600,000 student applications and counselor recommendation letters submitted via the Common App platform. We examine how the length and topical content of letters (e.g., sentences about Personal Qualities, Athletics, Intellectual Promise, etc.) relate to student self-identified race/ethnicity, sex, and proxies for socioeconomic status. Paired with regression analyses, we explore whether demographic differences in letter characteristics persist when accounting for additional student, school, and counselor characteristics, as well as among letters written by the same counselor and among students with comparably competitive standardized test scores. We ultimately find large and noteworthy naïve differences in letter length and content across nearly all demographic groups, many in alignment with known inequities (e.g., many more sentences about Athletics among White and higher-SES students, longer letters and more sentences on Personal Qualities for private school students). However, these differences vary drastically based on the exact controls and comparison groups included -- demonstrating that the ultimate implications of these letter differences for equity hinges on exactly how and when letters are used in admissions processes (e.g., among which groups of students are they used to "break ties"?). Findings do not point to a clear recommendation whether institutions should keep or discard letter requirements, but reflect the importance of reading letters and overall applications in the context of structural opportunity. We discuss additional implications and possible recommendations for college access and admissions policy/practice.

Ontologies are useful for automatic machine processing of domain knowledge as they represent it in a structured format. Yet, constructing ontologies requires substantial manual effort. To automate part of this process, large language models (LLMs) have been applied to solve various subtasks of ontology learning. However, this partial ontology learning does not capture the interactions between subtasks. We address this gap by introducing OLLM, a general and scalable method for building the taxonomic backbone of an ontology from scratch. Rather than focusing on subtasks, like individual relations between entities, we model entire subcomponents of the target ontology by finetuning an LLM with a custom regulariser that reduces overfitting on high-frequency concepts. We introduce a novel suite of metrics for evaluating the quality of the generated ontology by measuring its semantic and structural similarity to the ground truth. In contrast to standard metrics, our metrics use deep learning techniques to define more robust distance measures between graphs. Both our quantitative and qualitative results on Wikipedia show that OLLM outperforms subtask composition methods, producing more semantically accurate ontologies while maintaining structural integrity. We further demonstrate that our model can be effectively adapted to new domains, like arXiv, needing only a small number of training examples. Our source code and datasets are available at https://github.com/andylolu2/ollm.

In this paper, we study the problem of recovering a ground truth high dimensional piecewise linear curve $C^*(t):[0, 1]\to\mathbb{R}^d$ from a high noise Gaussian point cloud with covariance $\sigma^2I$ centered around the curve. We establish that the sample complexity of recovering $C^*$ from data scales with order at least $\sigma^6$. We then show that recovery of a piecewise linear curve from the third moment is locally well-posed, and hence $O(\sigma^6)$ samples is also sufficient for recovery. We propose methods to recover a curve from data based on a fitting to the third moment tensor with a careful initialization strategy and conduct some numerical experiments verifying the ability of our methods to recover curves. All code for our numerical experiments is publicly available on GitHub.

Integrated sensing and communication is regarded as a key enabler for next-generation wireless networks. To optimize the transmitted waveform for both sensing and communication, various performance metrics must be considered. This work focuses on sensing, and specifically on the mean square error (MSE) of channel estimation. Given the complexity of deriving the MSE, the Bayesian Cramer-Rao Bound (BCRB) is commonly recognized as a lower bound on the minimum MSE. However, the BCRB is not applicable to channels with discrete or mixed distributions. To address this limitation, a new lower bound based on a Poincar\'e inequality is proposed and applied to fading MIMO AWGN channels with blockage probability, and the behavior of the lower bound at high SNR is precisely characterized., Comment: Submitted to IEEE WCNC 2025

The widespread use of Large Language Models (LLMs) in software engineering has intensified the need for improved model and resource efficiency. In particular, for neural code generation, LLMs are used to translate function/method signature and DocString to executable code. DocStrings which capture user re quirements for the code and used as the prompt for LLMs, often contains redundant information. Recent advancements in prompt compression have shown promising results in Natural Language Processing (NLP), but their applicability to code generation remains uncertain. Our empirical study show that the state-of-the-art prompt compression methods achieve only about 10% reduction, as further reductions would cause significant performance degradation. In our study, we propose a novel compression method, ShortenDoc, dedicated to DocString compression for code generation. Our extensive experiments on six code generation datasets, five open-source LLMs (1B to 10B parameters), and one closed-source LLM GPT-4o confirm that ShortenDoc achieves 25-40% compression while preserving the quality of generated code, outperforming other baseline methods at similar compression levels. The benefit of this research is to improve efficiency and reduce the cost while maintaining the quality of the generated code, especially when calling third-party APIs, and is able to reduce the token processing cost by 25-40%., Comment: UNDER REVIEW

Brain imaging studies have demonstrated that diffusion MRI tractography geometric shape descriptors can inform the study of the brain's white matter pathways and their relationship to brain function. In this work, we investigate the possibility of utilizing a deep learning model to compute shape measures of the brain's white matter connections. We introduce a novel framework, TractShapeNet, that leverages a point cloud representation of tractography to compute five shape measures: length, span, volume, total surface area, and irregularity. We assess the performance of the method on a large dataset including 1065 healthy young adults. Experiments for shape measure computation demonstrate that our proposed TractShapeNet outperforms other point cloud-based neural network models in both the Pearson correlation coefficient and normalized error metrics. We compare the inference runtime results with the conventional shape computation tool DSI-Studio. Our results demonstrate that a deep learning approach enables faster and more efficient shape measure computation. We also conduct experiments on two downstream language cognition prediction tasks, showing that shape measures from TractShapeNet perform similarly to those computed by DSI-Studio. Our code will be available at: https://github.com/SlicerDMRI/TractShapeNet., Comment: 10 pages, 2 figures, 4 tables. This work has been submitted to the IEEE for possible publication

Reconstructing neuron morphology from 3D light microscope imaging data is critical to aid neuroscientists in analyzing brain networks and neuroanatomy. With the boost from deep learning techniques, a variety of learning-based segmentation models have been developed to enhance the signal-to-noise ratio of raw neuron images as a pre-processing step in the reconstruction workflow. However, most existing models directly encode the latent representative features of volumetric neuron data but neglect their intrinsic morphological knowledge. To address this limitation, we design a novel framework that distills the prior knowledge from a 2D Vision Transformer pre-trained on extensive 2D natural images to facilitate neuronal morphological learning of our 3D Vision Transformer. To bridge the knowledge gap between the 2D natural image and 3D microscopic morphologic domains, we propose a deformable tubular transferring strategy that adapts the pre-trained 2D natural knowledge to the inherent tubular characteristics of neuronal structure in the latent embedding space. The experimental results on the Janelia dataset of the BigNeuron project demonstrate that our method achieves a segmentation performance improvement of 4.53% in mean Dice and 3.56% in mean 95% Hausdorff distance., Comment: 9 pages, 3 figures, and 2 tables. This work has been submitted to the IEEE for possible publication

Egocentric vision captures the scene from the point of view of the camera wearer while exocentric vision captures the overall scene context. Jointly modeling ego and exo views is crucial to developing next-generation AI agents. The community has regained interest in the field of egocentric vision. While the third-person view and first-person have been thoroughly investigated, very few works aim to study both synchronously. Exocentric videos contain many relevant signals that are transferrable to egocentric videos. In this paper, we provide a broad overview of works combining egocentric and exocentric visions.

Magnet-superconductor hybrid (MSH) systems have recently emerged as one of the most significant developments in condensed matter physics. This has generated, in the last decade, a steadily rising interest in the understanding of their unique properties. They have been proposed as one of the most promising platforms for the establishment of topological superconductivity, which holds high potential for application in future quantum information technologies. Scanning tunneling microscopy (STM) and spectroscopy (STS) plays a crucial role in the race to unveil the fundamental origin of the unique properties of MSH systems, with the aim to discover new hybrid quantum materials capable of hosting topologically non-trivial unconventional superconducting phases. In particular, the combination of STM studies with tight-binding model calculations have represented, so far, the most successful approach to unveil and explain the emergent electronic properties of MSHs. The scope of this review is to offer a broad perspective on the field of MSHs from an atomic-level investigation point-of-view. The focus is on discussing the link between the magnetic ground state hosted by the hybrid system and the corresponding emergent superconducting phase. This is done for MSHs with both one-dimensional (atomic chains) and two-dimensional (atomic lattices and thin films) magnetic systems proximitized to conventional s-wave superconductors. We present a systematic categorization of the experimentally investigated systems with respect to defined experimentally accessible criteria to verify or falsify the presence of topological superconductivity and Majorana edge modes. Given the vast number of publications on the topic, we limit ourselves to discuss works which are most relevant to the search for topological superconductivity., Comment: Invited review article to appear in La Rivista del Nuovo Cimento

When conducting literature reviews, scientists often create literature review tables - tables whose rows are publications and whose columns constitute a schema, a set of aspects used to compare and contrast the papers. Can we automatically generate these tables using language models (LMs)? In this work, we introduce a framework that leverages LMs to perform this task by decomposing it into separate schema and value generation steps. To enable experimentation, we address two main challenges: First, we overcome a lack of high-quality datasets to benchmark table generation by curating and releasing arxivDIGESTables, a new dataset of 2,228 literature review tables extracted from ArXiv papers that synthesize a total of 7,542 research papers. Second, to support scalable evaluation of model generations against human-authored reference tables, we develop DecontextEval, an automatic evaluation method that aligns elements of tables with the same underlying aspects despite differing surface forms. Given these tools, we evaluate LMs' abilities to reconstruct reference tables, finding this task benefits from additional context to ground the generation (e.g. table captions, in-text references). Finally, through a human evaluation study we find that even when LMs fail to fully reconstruct a reference table, their generated novel aspects can still be useful., Comment: EMNLP 2024, 21 pages, 8 figures, 10 tables

We report a proof-of-principle realisation of a decoy-state BB84 QKD protocol with phase encoding over a record-breaking 17 km of MMF at a rate of 193 kbits/s, as well as over 1 Mbit/s at a distance of 1 km. These results suggest that QKD can be deployed over MMF in metropolitan-scale telecommunication connections. Such MMF metropolitan networks are ubiquitous - thus this advance could pave the way to wide scale metropolitan deployment. We also assess the advantages of adapting the OM3 channel using mode-matching photonic lanterns on the QBER, signal gain, and key rate and compare different encoding techniques in light of MMF propagation effects. This work confirms the suitability of current QKD technology for use in existing MMF links, unlocking new opportunities for quantum applications using legacy fibre., Comment: 9 pages, 6 figures, published in Optica Quantum

We present a new method for the verification of quantum circuits based on a novel symbolic representation of sets of quantum states using level-synchronized tree automata (LSTAs). LSTAs extend classical tree automata by labeling each transition with a set of choices, which are then used to synchronize subtrees of an accepted tree. Compared to the traditional tree automata, LSTAs have an incomparable expressive power while maintaining important properties, such as closure under union and intersection, and decidable language emptiness and inclusion. We have developed an efficient and fully automated symbolic verification algorithm for quantum circuits based on LSTAs. The complexity of supported gate operations is at most quadratic, dramatically improving the exponential worst-case complexity of an earlier tree automata-based approach. Furthermore, we show that LSTAs are a promising model for parameterized verification, i.e., verifying the correctness of families of circuits with the same structure for any number of qubits involved, which principally lies beyond the capabilities of previous automated approaches. We implemented this method as a C++ tool and compared it with three symbolic quantum circuit verifiers and two simulators on several benchmark examples. The results show that our approach can solve problems with sizes orders of magnitude larger than the state of the art.

We present the results of a search for gravitational-wave transients associated with core-collapse supernova SN 2023ixf, which was observed in the galaxy Messier 101 via optical emission on 2023 May 19th, during the LIGO-Virgo-KAGRA 15th Engineering Run. We define a five-day on-source window during which an accompanying gravitational-wave signal may have occurred. No gravitational waves have been identified in data when at least two gravitational-wave observatories were operating, which covered $\sim 14\%$ of this five-day window. We report the search detection efficiency for various possible gravitational-wave emission models. Considering the distance to M101 (6.7 Mpc), we derive constraints on the gravitational-wave emission mechanism of core-collapse supernovae across a broad frequency spectrum, ranging from 50 Hz to 2 kHz where we assume the GW emission occurred when coincident data are available in the on-source window. Considering an ellipsoid model for a rotating proto-neutron star, our search is sensitive to gravitational-wave energy $1 \times 10^{-5} M_{\odot} c^2$ and luminosity $4 \times 10^{-5} M_{\odot} c^2/\text{s}$ for a source emitting at 50 Hz. These constraints are around an order of magnitude more stringent than those obtained so far with gravitational-wave data. The constraint on the ellipticity of the proto-neutron star that is formed is as low as $1.04$, at frequencies above $1200$ Hz, surpassing results from SN 2019ejj., Comment: Main paper: 6 pages, 4 figures and 1 table. Total with appendices: 20 pages, 4 figures, and 1 table

We investigate coding schemes that map source symbols into multisets of an alphabet set. Such a formulation of source coding is an alternative approach to the traditional framework and is inspired by an object tracking problem over proximity sensor networks. We define a \textit{multiset combinatorial Gray code} as a mulitset code with fixed multiset cardinality that possesses combinatorial Gray code characteristic. For source codes that are organized as a grid, namely an integer lattice, we propose a solution by first constructing a mapping from the grid to the alphabet set, the codes are then defined as the images of rectangular blocks in the grid of fixed dimensions. We refer to the mapping as a \textit{color mapping} and the code as a \textit{color multiset code}. We propose the idea of product multiset code that enables us to construct codes for high dimensional grids based on 1-dimensional (1D) grids. We provide a detailed analysis of color multiset codes on 1D grids, focusing on codes that require the minimal number of colors. To illustrate the application of such a coding scheme, we consider an object tracking problem on 2D grids and show its efficiency, which comes from exploiting transmission parallelism. Some numerical results are presented to conclude the paper., Comment: 30 pages, 4 figures

The shape of the brain's white matter connections is relatively unexplored in diffusion MRI tractography analysis. While it is known that tract shape varies in populations and across the human lifespan, it is unknown if the variability in dMRI tractography-derived shape may relate to the brain's functional variability across individuals. This work explores the potential of leveraging tractography fiber cluster shape measures to predict subject-specific cognitive performance. We implement machine learning models to predict individual cognitive performance scores. We study a large-scale database from the HCP-YA study. We apply an atlas-based fiber cluster parcellation to the dMRI tractography of each individual. We compute 15 shape, microstructure, and connectivity features for each fiber cluster. Using these features as input, we train a total of 210 models to predict 7 different NIH Toolbox cognitive performance assessments. We apply an explainable AI technique, SHAP, to assess the importance of each fiber cluster for prediction. Our results demonstrate that shape measures are predictive of individual cognitive performance. The studied shape measures, such as irregularity, diameter, total surface area, volume, and branch volume, are as effective for prediction as microstructure and connectivity measures. The overall best-performing feature is a shape feature, irregularity, which describes how different a cluster's shape is from an idealized cylinder. Further interpretation using SHAP values suggest that fiber clusters with features highly predictive of cognitive ability are widespread throughout the brain, including fiber clusters from the superficial association, deep association, cerebellar, striatal, and projection pathways. This study demonstrates the strong potential of shape descriptors to enhance the study of the brain's white matter and its relationship to cognitive function.

Quantum networks (QNs) promise to enhance the performance of various quantum technologies in the near future by distributing entangled states over long distances. The first step towards this is to develop novel entanglement measures that are both informative and computationally tractable at large scales. While numerous such entanglement measures exist for discrete-variable (DV) systems, a comprehensive exploration for experimentally preferred continuous-variable (CV) systems is lacking. Here, we introduce a class of CV entanglement measures, among which we identify a non-convex entanglement monotone -- the ratio negativity, which possesses a simple, scalable form for characterizing the optimal entanglement swapping of pure Gaussian states. This characterization opens avenues for leveraging statistical physics tools to analyze swapping-protocol-based CV QNs.

The increasing complexity of large-scale FPGA accelerators poses significant challenges in achieving high performance while maintaining design productivity. High-level synthesis (HLS) has been adopted as a solution, but the mismatch between the high-level description and the physical layout often leads to suboptimal operating frequency. Although existing proposals for high-level physical synthesis, which use coarse-grained design partitioning, floorplanning, and pipelining to improve frequency, have gained traction, they lack a framework enabling (1) pipelining of real-world designs at arbitrary hierarchical levels, (2) integration of HLS blocks, vendor IPs, and handcrafted RTL designs, (3) portability to emerging new target FPGA devices, and (4) extensibility for the easy implementation of new design optimization tools. We present RapidStream IR, a practical high-level physical synthesis (HLPS) infrastructure for representing the composition of complex FPGA designs and exploring physical optimizations. Our approach introduces a flexible intermediate representation (IR) that captures interconnection protocols at arbitrary hierarchical levels, coarse-grained pipelining, and spatial information, enabling the creation of reusable passes for design frequency optimizations. RapidStream IR improves the frequency of a broad set of mixed-source designs by 7% to 62%, including large language models and genomics accelerators, and is portable to user-customizable new FPGA platforms. We further demonstrate its extensibility through case studies, showcasing the ability to facilitate future research.

In this paper, towards the unitary dual problem for classical groups, we propose a refinement of a conjecture of M. Tadi\'c: A representation of good parity is unitary if and only if it is of Arthur type; and a conjectural description of the whole unitary dual. We verify them for representations of corank at most $3$ for symplectic or split odd special orthogonal groups, based on Tadi\'c's classification. To this end, we develop new algorithms to determine whether a given irreducible representation is of Arthur type and give an inductive approach to classify the family of unitary representations that are of Arthur type. As applications, we carry out the explicit list on representations of Arthur type with corank at most $3$, and classify several new families of representations of Arthur type., Comment: Comments are welcome

Accurate classification of transients obtained from spectroscopic data are important to understand their nature and discover new classes of astronomical objects. For supernovae (SNe), SNID, NGSF (a Python version of SuperFit), and DASH are widely used in the community. Each tool provides its own metric to help determine classification, such as rlap of SNID, chi2/dof of NGSF, and Probability of DASH. However, we do not know how accurate these tools are, and they have not been tested with a large homogeneous dataset. Thus, in this work, we study the accuracy of these spectral classification tools using 4,646 SEDMachine spectra, which have accurate classifications obtained from the Zwicky Transient Facility Bright Transient Survey (BTS). Comparing our classifications with those from BTS, we have tested the classification accuracy in various ways. We find that NGSF has the best performance (overall Accuracy 87.6% when samples are split into SNe Ia and Non-Ia types), while SNID and DASH have similar performance with overall Accuracy of 79.3% and 76.2%, respectively. Specifically for SNe Ia, SNID can accurately classify them when rlap > 15 without contamination from other types, such as Ibc, II, SLSN, and other objects that are not SNe (Purity > 98%). For other types, determining their classification is often uncertain. We conclude that it is difficult to obtain an accurate classification from these tools alone. This results in additional human visual inspection effort being required in order to confirm the classification. To reduce this human visual inspection and to support the classification process for future large-scale surveys, this work provides supporting information, such as the accuracy of each tool as a function of its metric., Comment: 13 pages, 6 figures, and 6 tables; accepted for publication in PASP

The use of Large Language Models (LLMs) in automated test generation is gaining popularity, with much of the research focusing on metrics like compilability rate, code coverage and bug detection. However, an equally important quality metric is the presence of test smells design flaws or anti patterns in test code that hinder maintainability and readability. In this study, we explore the diffusion of test smells in LLM generated unit test suites and compare them to those found in human written ones. We analyze a benchmark of 20,500 LLM-generated test suites produced by four models (GPT-3.5, GPT-4, Mistral 7B, and Mixtral 8x7B) across five prompt engineering techniques, alongside a dataset of 780,144 human written test suites from 34,637 projects. Leveraging TsDetect, a state of the art tool capable of detecting 21 different types of test smells, we identify and analyze the prevalence and co-occurrence of various test smells in both human written and LLM-generated test suites. Our findings reveal new insights into the strengths and limitations of LLMs in test generation. First, regarding prevalence, we observe that LLMs frequently generate tests with common test smells, such as Magic Number Test and Assertion Roulette. Second, in terms of co occurrence, certain smells, like Long Test and Useless Test, tend to co occur in LLM-generated suites, influenced by specific prompt techniques. Third, we find that project complexity and LLM specific factors, including model size and context length, significantly affect the prevalence of test smells. Finally, the patterns of test smells in LLM-generated tests often mirror those in human-written tests, suggesting potential data leakage from training datasets. These insights underscore the need to refine LLM-based test generation for cleaner code and suggest improvements in both LLM capabilities and software testing practices.

Bayesian optimization (BO) has become an indispensable tool for autonomous decision-making across diverse applications from autonomous vehicle control to accelerated drug and materials discovery. With the growing interest in self-driving laboratories, BO of chemical systems is crucial for machine learning (ML) guided experimental planning. Typically, BO employs a regression surrogate model to predict the distribution of unseen parts of the search space. However, for the selection of molecules, picking the top candidates with respect to a distribution, the relative ordering of their properties may be more important than their exact values. In this paper, we introduce Rank-based Bayesian Optimization (RBO), which utilizes a ranking model as the surrogate. We present a comprehensive investigation of RBO's optimization performance compared to conventional BO on various chemical datasets. Our results demonstrate similar or improved optimization performance using ranking models, particularly for datasets with rough structure-property landscapes and activity cliffs. Furthermore, we observe a high correlation between the surrogate ranking ability and BO performance, and this ability is maintained even at early iterations of BO optimization when using ranking surrogate models. We conclude that RBO is an effective alternative to regression-based BO, especially for optimizing novel chemical compounds., Comment: 14 + 4 pages, 5 + 3 figures

The magnetar SGR 1935+2154 is the only known Galactic source of fast radio bursts (FRBs). FRBs from SGR 1935+2154 were first detected by CHIME/FRB and STARE2 in 2020 April, after the conclusion of the LIGO, Virgo, and KAGRA Collaborations' O3 observing run. Here we analyze four periods of gravitational wave (GW) data from the GEO600 detector coincident with four periods of FRB activity detected by CHIME/FRB, as well as X-ray glitches and X-ray bursts detected by NICER and NuSTAR close to the time of one of the FRBs. We do not detect any significant GW emission from any of the events. Instead, using a short-duration GW search (for bursts $\leq$ 1 s) we derive 50\% (90\%) upper limits of $10^{48}$ ($10^{49}$) erg for GWs at 300 Hz and $10^{49}$ ($10^{50}$) erg at 2 kHz, and constrain the GW-to-radio energy ratio to $\leq 10^{14} - 10^{16}$. We also derive upper limits from a long-duration search for bursts with durations between 1 and 10 s. These represent the strictest upper limits on concurrent GW emission from FRBs., Comment: 15 pages of text including references, 4 figures, 5 tables

The High Energy Modular Array (HEMA) is one of three instruments that compose the STROBE-X mission concept. The HEMA is a large-area, high-throughput non-imaging pointed instrument based on the Large Area Detector developed as part of the LOFT mission concept. It is designed for spectral timing measurements of a broad range of sources and provides a transformative increase in sensitivity to X-rays in the energy range of 2--30 keV compared to previous instruments, with an effective area of 3.4 m$^{2}$ at 8.5 keV and an energy resolution of better than 300 eV at 6 keV in its nominal field of regard., Comment: 16 pages, 10 figures

The first very long baseline interferometry (VLBI) detections at 870$\mu$m wavelength (345$\,$GHz frequency) are reported, achieving the highest diffraction-limited angular resolution yet obtained from the surface of the Earth, and the highest-frequency example of the VLBI technique to date. These include strong detections for multiple sources observed on inter-continental baselines between telescopes in Chile, Hawaii, and Spain, obtained during observations in October 2018. The longest-baseline detections approach 11$\,$G$\lambda$ corresponding to an angular resolution, or fringe spacing, of 19$\mu$as. The Allan deviation of the visibility phase at 870$\mu$m is comparable to that at 1.3$\,$mm on the relevant integration time scales between 2 and 100$\,$s. The detections confirm that the sensitivity and signal chain stability of stations in the Event Horizon Telescope (EHT) array are suitable for VLBI observations at 870$\mu$m. Operation at this short wavelength, combined with anticipated enhancements of the EHT, will lead to a unique high angular resolution instrument for black hole studies, capable of resolving the event horizons of supermassive black holes in both space and time., Comment: Corresponding author: S. Doeleman

We introduce Pixtral-12B, a 12--billion-parameter multimodal language model. Pixtral-12B is trained to understand both natural images and documents, achieving leading performance on various multimodal benchmarks, surpassing a number of larger models. Unlike many open-source models, Pixtral is also a cutting-edge text model for its size, and does not compromise on natural language performance to excel in multimodal tasks. Pixtral uses a new vision encoder trained from scratch, which allows it to ingest images at their natural resolution and aspect ratio. This gives users flexibility on the number of tokens used to process an image. Pixtral is also able to process any number of images in its long context window of 128K tokens. Pixtral 12B substanially outperforms other open models of similar sizes (Llama-3.2 11B \& Qwen-2-VL 7B). It also outperforms much larger open models like Llama-3.2 90B while being 7x smaller. We further contribute an open-source benchmark, MM-MT-Bench, for evaluating vision-language models in practical scenarios, and provide detailed analysis and code for standardized evaluation protocols for multimodal LLMs. Pixtral-12B is released under Apache 2.0 license.

With the increasing use of neural policies in control systems, ensuring their safety and reliability has become a critical software engineering task. One prevalent approach to ensuring the safety of neural policies is to deploy programmatic runtime shields alongside them to correct their unsafe commands. However, the programmatic runtime shields synthesized by existing methods are either computationally expensive or insufficiently permissive, resulting in high overhead and unnecessary interventions on the system. To address these challenges, we propose Aegis, a novel framework that synthesizes lightweight and permissive programmatic runtime shields for neural policies. Aegis achieves this by formulating the seeking of a runtime shield as a sketch-based program synthesis problem and proposing a novel method that leverages counterexample-guided inductive synthesis and Bayesian optimization to solve it. To evaluate Aegis and its synthesized shields, we use four representative control systems and compare Aegis with the current state-of-the-art. Our results show that the programmatic runtime shields synthesized by Aegis can correct all unsafe commands from neural policies, ensuring that the systems do not violate any desired safety properties at all times. Compared to the current state-of-the-art, Aegis's shields exhibit a 2.1$\times$ reduction in time overhead and a 4.4$\times$ reduction in memory usage, suggesting that they are much more lightweight. Moreover, Aegis's shields incur an average of 1.6$\times$ fewer interventions than other shields, showing better permissiveness., Comment: Under Review by ACM Transactions on Software Engineering and Methodology (TOSEM)

Given the increasing adoption of modern AI-enabled control systems, ensuring their safety and reliability has become a critical task in software testing. One prevalent approach to testing control systems is falsification, which aims to find an input signal that causes the control system to violate a formal safety specification using optimization algorithms. However, applying falsification to AI-enabled control systems poses two significant challenges: (1)~it requires the system to execute numerous candidate test inputs, which can be time-consuming, particularly for systems with AI models that have many parameters, and (2)~multiple safety requirements are typically defined as a conjunctive specification, which is difficult for existing falsification approaches to comprehensively cover. This paper introduces Synthify, a falsification framework tailored for AI-enabled control systems. Our approach performs falsification in a two-phase process. At the start, Synthify synthesizes a program that implements one or a few linear controllers to serve as a proxy for the AI controller. This proxy program mimics the AI controller's functionality but is computationally more efficient. Then, Synthify employs the $\epsilon$-greedy strategy to sample a promising sub-specification from the conjunctive safety specification. It then uses a Simulated Annealing-based falsification algorithm to find violations of the sampled sub-specification for the control system. To evaluate Synthify, we compare it to PSY-TaLiRo, a state-of-the-art and industrial-strength falsification tool, on 8 publicly available control systems. On average, Synthify achieves a 83.5% higher success rate in falsification compared to PSY-TaLiRo with the same budget of falsification trials. The safety violations found by Synthify are also more diverse than those found by PSY-TaLiRo, covering 137.7% more sub-specifications., Comment: Under Review by ACM Transactions on Software Engineering and Methodology (TOSEM)

We present a comprehensive theoretical investigation of the strain-modulated excitonic properties of uni-axially strained transition-metal dichalcogenide monolayers (TMD-MLs) by solving the Bethe-Salpeter equation (BSE) established on the basis of first principles. We show that imposing an uni-axial strain onto a MoS_$2$ monolayers leads to the diversified band dispersions of the bright exciton (BX), gray exciton (GX), and dark exciton (DX) states, as a consequence of the competitive interplay between strain-induced valley drift (VD) and momentum-dependent electron-hole exchange interaction (EHEI). While the band dispersions of BX doublet in the light-accessible small reciprocal area remain almost unchanged against strain, the band dispersion of DX is reshaped by an increasing uni-axial strain from a parabola to a Mexican-hat-like profile, featured with unusual sign-reversal of the heavy effective mass and strain-activated brightness. In contrast, the effective mass of GX is drastically lightened by uni-axial strain and remains always positive. We show that the strain-diversified exciton band dispersions leads to the distinct exciton diffusivities and angle-resolved optical patterns of BX, GX, and DX in a strained TMD-ML, suggesting the feasibility of {\it spatially} resolving spinallowed and -forbidden excitons in exciton transport experiments and angle-resolved optical spectroscopies., Comment: 18pages, 5 figures

In this paper, we give a definition of weak stability condition on a triangulated category. The difference between our definition and existing definitions is that we allow objects in the kernel to have non-maximal phases. We then construct four types of weak stability conditions that naturally occur on Weierstrass ellitpic surfaces as limites of Bridgeland stability conditions., Comment: The preprint arXiv:2306.05620v1 has been divided into two papers; this is Part 1 of arXiv:2306.05620v1

Large language models (LLMs) have been shown to exhibit emergent abilities in some downstream tasks, where performance seems to stagnate at first and then improve sharply and unpredictably with scale beyond a threshold. By dividing questions in the datasets according to difficulty level by average performance, we observe U-shaped scaling for hard questions, and inverted-U scaling followed by steady improvement for easy questions. Moreover, the emergence threshold roughly coincides with the point at which performance on easy questions reverts from inverse scaling to standard scaling. Capitalizing on the observable though opposing scaling trend on easy and hard questions, we propose a simple yet effective pipeline, called Slice-and-Sandwich, to predict both the emergence threshold and model performance beyond the threshold., Comment: Preprint. Under review

Barnard's star is a primary target within the ESPRESSO guaranteed time observations (GTO) as it is the second closest neighbour to our Sun after the $\alpha$ Centauri stellar system. We present here a large set of 156 ESPRESSO observations of Barnard's star carried out over four years with the goal of exploring periods of shorter than 50 days, thus including the habitable zone (HZ). Our analysis of ESPRESSO data using Gaussian process (GP) to model stellar activity suggests a long-term activity cycle at 3200d and confirms stellar activity due to rotation at 140d as the dominant source of radial velocity (RV) variations. These results are in agreement with findings based on publicly available HARPS, HARPS-N, and CARMENES data. ESPRESSO RVs do not support the existence of the previously reported candidate planet at 233d. After subtracting the GP model, ESPRESSO RVs reveal several short-period candidate planet signals at periods of 3.15d, 4.12d, 2.34d, and 6.74d. We confirm the 3.15d signal as a sub-Earth mass planet, with a semi-amplitude of $55 \pm 7$cm/s, leading to a planet minimum mass $m_p \sin i$ of $0.37 \pm 0.05$Mearth, which is about three times the mass of Mars. ESPRESSO RVs suggest the possible existence of a candidate system with four sub-Earth mass planets in circular orbits with semi-amplitudes from 20 to 47cm/s, thus corresponding to minimum masses in the range of 0.17-0.32Mearth. The sub-Earth mass planet at $3.1533 \pm 0.0006$d is in a close-to circular orbit with a semi-major axis of $0.0229 \pm 0.0003$AU, thus located inwards from the HZ of Barnard's star, with an equilibrium temperature of 400K. Additional ESPRESSO observations would be required to confirm that the other three candidate signals originate from a compact short-period planet system orbiting Barnard's star inwards from its HZ., Comment: Accepted for publication in Astronomy and Astrophysics

Today's most advanced multimodal models remain proprietary. The strongest open-weight models rely heavily on synthetic data from proprietary VLMs to achieve good performance, effectively distilling these closed models into open ones. As a result, the community is still missing foundational knowledge about how to build performant VLMs from scratch. We present Molmo, a new family of VLMs that are state-of-the-art in their class of openness. Our key innovation is a novel, highly detailed image caption dataset collected entirely from human annotators using speech-based descriptions. To enable a wide array of user interactions, we also introduce a diverse dataset mixture for fine-tuning that includes in-the-wild Q&A and innovative 2D pointing data. The success of our approach relies on careful choices for the model architecture details, a well-tuned training pipeline, and, most critically, the quality of our newly collected datasets, all of which will be released. The best-in-class 72B model within the Molmo family not only outperforms others in the class of open weight and data models but also compares favorably against proprietary systems like GPT-4o, Claude 3.5, and Gemini 1.5 on both academic benchmarks and human evaluation. We will be releasing all of our model weights, captioning and fine-tuning data, and source code in the near future. Select model weights, inference code, and demo are available at https://molmo.allenai.org.

LLMs can now act as autonomous agents that interact with digital environments and complete specific objectives (e.g., arranging an online meeting). However, accuracy is still far from satisfactory, partly due to a lack of large-scale, direct demonstrations for digital tasks. Obtaining supervised data from humans is costly, and automatic data collection through exploration or reinforcement learning relies on complex environmental and content setup, resulting in datasets that lack comprehensive coverage of various scenarios. On the other hand, there is abundant knowledge that may indirectly assist task completion, such as online tutorials that were created for human consumption. In this work, we present Synatra, an approach that effectively transforms this indirect knowledge into direct supervision at scale. We define different types of indirect knowledge, and carefully study the available sources to obtain it, methods to encode the structure of direct demonstrations, and finally methods to transform indirect knowledge into direct demonstrations. We use 100k such synthetically-created demonstrations to finetune a 7B CodeLlama, and demonstrate that the resulting agent surpasses all comparably sized models on three web-based task benchmarks Mind2Web, MiniWoB++ and WebArena, as well as surpassing GPT-3.5 on WebArena and Mind2Web. In addition, while synthetic demonstrations prove to be only 3% the cost of human demonstrations (at $0.031 each), we show that the synthetic demonstrations can be more effective than an identical number of human demonstrations collected from limited domains.

We conducted controlled experimental bias audits for four versions of ChatGPT, which we asked to recommend an opening offer in salary negotiations for a new hire. We submitted 98,800 prompts to each version, systematically varying the employee's gender, university, and major, and tested prompts in voice of each side of the negotiation: the employee versus employer. We find ChatGPT as a multi-model platform is not robust and consistent enough to be trusted for such a task. We observed statistically significant salary offers when varying gender for all four models, although with smaller gaps than for other attributes tested. The largest gaps were different model versions and between the employee- vs employer-voiced prompts. We also observed substantial gaps when varying university and major, but many of the biases were not consistent across model versions. We tested for fictional and fraudulent universities and found wildly inconsistent results across cases and model versions. We make broader contributions to the AI/ML fairness literature. Our scenario and our experimental design differ from mainstream AI/ML auditing efforts in key ways. Bias audits typically test discrimination for protected classes like gender, which we contrast with testing non-protected classes of university and major. Asking for negotiation advice includes how aggressive one ought to be in a negotiation relative to known empirical salary distributions and scales, which is a deeply contextual and personalized task that has no objective ground truth to validate. These results raise concerns for the specific model versions we tested and ChatGPT as a multi-model platform in continuous development. Our epistemology does not permit us to definitively certify these models as either generally biased or unbiased on the attributes we test, but our study raises matters of concern for stakeholders to further investigate.

Ultra hot Jupiters (gas giants, Teq>2000 K) are intriguing exoplanets due to their extreme atmospheres. Their torrid daysides can be characterised using ground-based high-resolution emission spectroscopy. We search for signatures of neutral and singly ionised iron (Fe I and Fe II) in the dayside of the ultra hot Jupiter WASP-76 b, as these species were detected via transmission spectroscopy in this exoplanet. Furthermore, we aim to confirm the existence of a thermal inversion layer, which has been reported in previous studies, and attempt to constrain its properties. We observed WASP-76 b on four epochs with ESPRESSO at the VLT, at orbital phases shortly before and after the secondary transit, when the dayside is in view. We present the first analysis of high-resolution optical emission spectra for this exoplanet. We compare the data to synthetic templates from petitRADTRANS, using cross-correlation function techniques. We detect a blueshifted (-4.7+-0.3 km/s) Fe I emission signature on the dayside of WASP-76 b at 6.0-sigma. The signal is detected independently both before and after the eclipse, and blueshifted in both cases. The presence of iron emission features confirms the existence of a thermal inversion layer. Fe II was not detected, possibly because this species is located in the upper layers of the atmosphere, which are more optically thin. Thus the Fe II signature on the dayside of WASP-76 b is too weak to be detected with emission spectroscopy. We propose that the blueshifted Fe I signature is created by material rising from the hot spot to the upper layers of the atmosphere, and discuss possible scenarios related to the position of the hotspot. This work unveils some of the dynamic processes ongoing on the dayside of WASP-76 b through the analysis of the Fe I signature from its atmosphere, and complements previous knowledge obtained from transmission studies., Comment: 15 pages, 8 figures + 2 figures in appendix. Abstract slightly abridged. Published in Astronomy & Astrophysics

Reservoir computing (RC) is an effective method for predicting chaotic systems by using a high-dimensional dynamic reservoir with fixed internal weights, while keeping the learning phase linear, which simplifies training and reduces computational complexity compared to fully trained recurrent neural networks (RNNs). Quantum reservoir computing (QRC) uses the exponential growth of Hilbert spaces in quantum systems, allowing for greater information processing, memory capacity, and computational power. However, the original QRC proposal requires coherent injection of inputs multiple times, complicating practical implementation. We present a hybrid quantum-classical approach that implements memory through classical post-processing of quantum measurements. This approach avoids the need for multiple coherent input injections and is evaluated on benchmark tasks, including the chaotic Mackey-Glass time series prediction. We tested our model on two physical platforms: a fully connected Ising model and a Rydberg atom array. The optimized model demonstrates promising predictive capabilities, achieving a higher number of steps compared to previously reported approaches., Comment: 8 pages, 7 figures

Memristive systems exhibit dynamics that depend on their past states, making them useful as memory units. Due to their functional resemblance to real synapses, memristors can also constitute the building blocks of neural networks for a variety of tasks. Recently, quantum memristor models have been proposed and notably, a photonic quantum memristor (PQM) has been experimentally proven. In this work, we explore and characterize various quantum properties that emerge from this specific model of PQM. Firstly, we find that a single PQM displays memristive dynamics on its quantum coherence. Secondly, we show that a network made of two independent PQMs can manifest memory effects on the dynamics of both entanglement and coherence of correlated photons traveling through the network, regardless of their distance. Additionally, we build and run a circuit-model of the PQM on a real qubit-based quantum computer (IBM-Q), showing that: (i) this system can effectively be used for non-linear quantum computing under specific conditions, and (ii) digital quantum simulations can reproduce the dynamics of a memristive quantum system in a non-Markovian regime., Comment: 9 pages, 7 figures

Selecting the best code solution from multiple generated ones is an essential task in code generation, which can be achieved by using some reliable validators (e.g., developer-written test cases) for assistance. Since reliable test cases are not always available and can be expensive to build in practice, researchers propose to automatically generate test cases to assess code solutions. However, when both code solutions and test cases are plausible and not reliable, selecting the best solution becomes challenging. Although some heuristic strategies have been proposed to tackle this problem, they lack a strong theoretical guarantee and it is still an open question whether an optimal selection strategy exists. Our work contributes in two ways. First, we show that within a Bayesian framework, the optimal selection strategy can be defined based on the posterior probability of the observed passing states between solutions and tests. The problem of identifying the best solution is then framed as an integer programming problem. Second, we propose an efficient approach for approximating this optimal (yet uncomputable) strategy, where the approximation error is bounded by the correctness of prior knowledge. We then incorporate effective prior knowledge to tailor code generation tasks. Both theoretical and empirical studies confirm that existing heuristics are limited in selecting the best solutions with plausible test cases. Our proposed approximated optimal strategy B4 significantly surpasses existing heuristics in selecting code solutions generated by large language models (LLMs) with LLM-generated tests, achieving a relative performance improvement by up to 50% over the strongest heuristic and 246% over the random selection in the most challenging scenarios. Our code is publicly available at https://github.com/ZJU-CTAG/B4., Comment: accepted by ASE' 24 (full paper)

On the 29th of May 2023, the LIGO-Virgo-KAGRA Collaboration observed a compact binary coalescence event consistent with a neutron star-black hole merger, though the heavier object of mass 2.5-4.5 $M_\odot$ would fall into the purported lower mass gap. An alternative explanation for apparent observations of events in this mass range has been suggested as strongly gravitationally lensed binary neutron stars. In this scenario, magnification would lead to the source appearing closer and heavier than it really is. Here, we investigate the chances and possible consequences for the GW230529 event to be gravitationally lensed. We find this would require high magnifications and we obtain low rates for observing such an event, with a relative fraction of lensed versus unlensed observed events of $2 \times 10^{-3}$ at most. When comparing the lensed and unlensed hypotheses accounting for the latest rates and population model, we find a 1/58 chance of lensing, disfavoring this option. Moreover, when the magnification is assumed to be strong enough to bring the mass of the heavier binary component below the standard limits on neutron star masses, we find high probability for the lighter object to have a sub-solar mass, making the binary even more exotic than a mass-gap neutron star-black hole system. Even when the secondary is not sub-solar, its tidal deformability would likely be measurable, which is not the case for GW230529. Finally, we do not find evidence for extra lensing signatures such as the arrival of additional lensed images, type-II image dephasing, or microlensing. Therefore, we conclude it is unlikely for GW230529 to be a strongly gravitationally lensed binary neutron star signal., Comment: 15 pages, 11 figures

Second language (L2) learners can improve their pronunciation by imitating golden speech, especially when the speech that aligns with their respective speech characteristics. This study explores the hypothesis that learner-specific golden speech generated with zero-shot text-to-speech (ZS-TTS) techniques can be harnessed as an effective metric for measuring the pronunciation proficiency of L2 learners. Building on this exploration, the contributions of this study are at least two-fold: 1) design and development of a systematic framework for assessing the ability of a synthesis model to generate golden speech, and 2) in-depth investigations of the effectiveness of using golden speech in automatic pronunciation assessment (APA). Comprehensive experiments conducted on the L2-ARCTIC and Speechocean762 benchmark datasets suggest that our proposed modeling can yield significant performance improvements with respect to various assessment metrics in relation to some prior arts. To our knowledge, this study is the first to explore the role of golden speech in both ZS-TTS and APA, offering a promising regime for computer-assisted pronunciation training (CAPT)., Comment: 11 pages, 4 figures, 4 tables

Automated speaking assessment in conversation tests (ASAC) aims to evaluate the overall speaking proficiency of an L2 (second-language) speaker in a setting where an interlocutor interacts with one or more candidates. Although prior ASAC approaches have shown promising performance on their respective datasets, there is still a dearth of research specifically focused on incorporating the coherence of the logical flow within a conversation into the grading model. To address this critical challenge, we propose a hierarchical graph model that aptly incorporates both broad inter-response interactions (e.g., discourse relations) and nuanced semantic information (e.g., semantic words and speaker intents), which is subsequently fused with contextual information for the final prediction. Extensive experimental results on the NICT-JLE benchmark dataset suggest that our proposed modeling approach can yield considerable improvements in prediction accuracy with respect to various assessment metrics, as compared to some strong baselines. This also sheds light on the importance of investigating coherence-related facets of spoken responses in ASAC., Comment: Accepted by IEEE SLT 2024

Generative adversarial network (GAN) is a strong deep learning model that has shown its value in practical applications such as image processing and data enhancement. Here, we propose a quantum topological code decoder based on GAN and we apply it to optimize the fault-tolerant quantum teleportation system. We construct the generator and discriminator networks of GAN, train the network using the eigenvalue dataset of the topological code, and obtain an optimized decoder with high decoding threshold. The decoding experiments at code distances $d=3$ and $d=5$ show that the error correction success rate of this model reaches 99.895\%. In the experiment, the fidelity threshold of this GAN decoder is about $P=0.2108$, which is significantly improved compared with the threshold $P=0.1099$ of the classical decoding model. In addition, the quantum teleportation system, optimized for noise resistance under $d=3$ topological code, shows a noticeable fidelity improvement within the non-polarized noise threshold range of $P<0.06503$, while under $d=5$ topological code optimization, there is a significant fidelity improvement within the non-polarized noise threshold range of $P<0.07512$. The proposed GAN model supplies a novel approach for topological code decoders and its principles can be applied to different kinds of noise processing., Comment: 12 pages, 11 figures

Quantum extreme learning machines (QELMs) leverage untrained quantum dynamics to efficiently process information encoded in input quantum states, avoiding the high computational cost of training more complicated nonlinear models. On the other hand, quantum information scrambling (QIS) quantifies how the spread of quantum information into correlations makes it irretrievable from local measurements. Here, we explore the tight relation between QIS and the predictive power of QELMs. In particular, we show efficient state estimation is possible even beyond the scrambling time, for many different types of dynamics -- in fact, we show that in all the cases we studied, the reconstruction efficiency at long interaction times matches the optimal one offered by random global unitary dynamics. These results offer promising venues for robust experimental QELM-based state estimation protocols, as well as providing novel insights into the nature of QIS from a state estimation perspective.

We present the first homogeneous release of several thousand Type Ia supernovae (SNe Ia), all having spectroscopic classification, and spectroscopic redshifts for half the sample. This release, named the "DR2", contains 3628 nearby (z < 0.3) SNe Ia discovered, followed and classified by the Zwicky Transient Facility survey between March 2018 and December 2020. Of these, 3000 have good-to-excellent sampling and 2667 pass standard cosmology light-curve quality cuts. This release is thus the largest SN Ia release to date, increasing by an order of magnitude the number of well characterized low-redshift objects. With the "DR2", we also provide a volume-limited (z < 0.06) sample of nearly a thousand SNe Ia. With such a large, homogeneous and well controlled dataset, we are studying key current questions on SN cosmology, such as the linearity SNe Ia standardization, the SN and host dependencies, the diversity of the SN Ia population, and the accuracy of the current light-curve modeling. These, and more, are studied in detail in a series of articles associated with this release. Alongside the SN Ia parameters, we publish our force-photometry gri-band light curves, 5138 spectra, local and global host properties, observing logs, and a python tool to ease use and access of these data. The photometric accuracy of the "DR2" is not yet suited for cosmological parameter inference, which will follow as "DR2.5" release. We nonetheless demonstrate that the multi-thousand SN Ia Hubble Diagram has a typical 0.15 mag scatter., Comment: ZTF SN Ia DR2 release paper. Submitted to A&A (ZTF DR2 Special Issue). Already 1 response to referee

In this work, we report a pronounced light upconversion in few-layer transition metal dichalcogenides. Our joint theory-experiment study attributes the upconversion photoluminescence to a resonant exciton-exciton annihilation involving a pair of dark excitons with opposite momenta, followed by the spontaneous emission of upconverted bright excitons, which can have a high upconversion efficiency. Additionally, the upconversion photoluminescence is generic in MoS2, MoSe2, WS2, and WSe2, showing a high tuneability from green to ultraviolet light.

Information retrieval methods often rely on a single embedding model trained on large, general-domain datasets like MSMARCO. While this approach can produce a retriever with reasonable overall performance, models trained on domain-specific data often yield better results within their respective domains. While prior work in information retrieval has tackled this through multi-task training, the topic of combining multiple domain-specific expert retrievers remains unexplored, despite its popularity in language model generation. In this work, we introduce RouterRetriever, a retrieval model that leverages multiple domain-specific experts along with a routing mechanism to select the most appropriate expert for each query. It is lightweight and allows easy addition or removal of experts without additional training. Evaluation on the BEIR benchmark demonstrates that RouterRetriever outperforms both MSMARCO-trained (+2.1 absolute nDCG@10) and multi-task trained (+3.2) models. This is achieved by employing our routing mechanism, which surpasses other routing techniques (+1.8 on average) commonly used in language modeling. Furthermore, the benefit generalizes well to other datasets, even in the absence of a specific expert on the dataset. To our knowledge, RouterRetriever is the first work to demonstrate the advantages of using multiple domain-specific expert embedding models with effective routing over a single, general-purpose embedding model in retrieval tasks.

We introduce OLMoE, a fully open, state-of-the-art language model leveraging sparse Mixture-of-Experts (MoE). OLMoE-1B-7B has 7 billion (B) parameters but uses only 1B per input token. We pretrain it on 5 trillion tokens and further adapt it to create OLMoE-1B-7B-Instruct. Our models outperform all available models with similar active parameters, even surpassing larger ones like Llama2-13B-Chat and DeepSeekMoE-16B. We present various experiments on MoE training, analyze routing in our model showing high specialization, and open-source all aspects of our work: model weights, training data, code, and logs., Comment: 61 pages (24 main), 36 figures, 14 tables

This article presents a new approach based on MiniRocket, called SelF-Rocket, for fast time series classification (TSC). Unlike existing approaches based on random convolution kernels, it dynamically selects the best couple of input representations and pooling operator during the training process. SelF-Rocket achieves state-of-the-art accuracy on the University of California Riverside (UCR) TSC benchmark datasets.

With the advancement of Large Language Models (LLMs), their application in Software Quality Assurance (SQA) has increased. However, the current focus of these applications is predominantly on ChatGPT. There remains a gap in understanding the performance of various LLMs in this critical domain. This paper aims to address this gap by conducting a comprehensive investigation into the capabilities of several LLMs across two SQA tasks: fault localization and vulnerability detection. We conducted comparative studies using GPT-3.5, GPT-4o, and four other publicly available LLMs (LLaMA-3-70B, LLaMA-3-8B, Gemma-7B, and Mixtral-8x7B), to evaluate their effectiveness in these tasks. Our findings reveal that several LLMs can outperform GPT-3.5 in both tasks. Additionally, even the lower-performing LLMs provided unique correct predictions, suggesting the potential of combining different LLMs' results to enhance overall performance. By implementing a voting mechanism to combine the LLMs' results, we achieved more than a 10% improvement over the GPT-3.5 in both tasks. Furthermore, we introduced a cross-validation approach to refine the LLM answer by validating one LLM answer against another using a validation prompt. This approach led to performance improvements of 16% in fault localization and 12% in vulnerability detection compared to the GPT-3.5, with a 4% improvement compared to the best-performed LLMs. Our analysis also indicates that the inclusion of explanations in the LLMs' results affects the effectiveness of the cross-validation technique.

Interest in quantum machine learning is increasingly growing due to its potential to offer more efficient solutions for problems that are difficult to tackle with classical methods. In this context, the research work presented here focuses on the use of quantum machine learning techniques for image classification tasks. We exploit a quantum extreme learning machine by taking advantage of its rich feature map provided by the quantum reservoir substrate. We systematically analyse different phases of the quantum extreme learning machine process, from the dataset preparation to the image final classification. In particular, we have tested different encodings, together with Principal Component Analysis, the use of Auto-Encoders, as well as the dynamics of the model through the use of different Hamiltonians for the quantum reservoir. Our results show that the introduction of a quantum reservoir systematically improves the accuracy of the classifier. Additionally, while different encodings can lead to significantly different performances, Hamiltonians with varying degrees of connectivity exhibit the same discrimination rate, provided they are interacting., Comment: 11 pages, 6 figures