Search

Several algorithms implemented by language models have recently been successfully reversed-engineered. However, these findings have been concentrated on specific tasks and models, leaving it unclear how universal circuits are across different settings. In this paper, we study the circuits implemented by Gemma 2B for solving the subject-verb agreement task across two different languages, English and Spanish. We discover that both circuits are highly consistent, being mainly driven by a particular attention head writing a `subject number' signal to the last residual stream, which is read by a small set of neurons in the final MLPs. Notably, this subject number signal is represented as a direction in the residual stream space, and is language-independent. We demonstrate that this direction has a causal effect on the model predictions, effectively flipping the Spanish predicted verb number by intervening with the direction found in English. Finally, we present evidence of similar behavior in other models within the Gemma 1 and Gemma 2 families., Comment: Accepted at EMNLP 2024 Findings

Left-Right Models (LRMs) are one of the most relevant extensions of the Standard Model (SM) of particle physics. They introduce an extended gauge sector and can restore parity (P) or charge conjugation (C) symmetries at high enough energies. These theories can be embedded in other more fundamental ones with larger gauge groups. Consequently, the restoration of the C or P symmetries can be pushed towards higher energy scales compared to the scale of the Spontaneous Symmetry Breaking (SSB) of the LRM gauge group. We study three LRMs with different specific realizations of the scalar sector without imposing any additional discrete symmetry on the theory. We present bounds on the masses of the new gauge bosons using data from the LHC Run 2 and study rare meson decays, discussing the structure of the right-handed quark mixing matrix and the impact of the neutrino and scalar sectors. Collider bounds valid for specific LRM realizations are alleviated bringing New Physics (NP) effects in flavour observables closer to an observable level., Comment: Contribution to the Beyond the Standard Model Session of the 42nd International Conference on High Energy Physics

The necessary and sufficient conditions for a perfect fluid solution to define a spatially-homogeneous cosmology are achieved. These conditions are Intrinsic, Deductive, Explicit and ALgorithmic, and they offer an IDEAL labeling of these geometries. When a three-dimensional group acts on the three-dimensional space-like orbits, the Bianchi type of the model is also obtained., Comment: 25 pages, 4 figures, 1 supplementary documentation

The thermodynamic interpretation of the Stephani Universes is studied in detail. The general expression of the speed of sound and of the thermodynamic schemes associated with a thermodynamic solution is obtained. The constraints imposed on the solutions by considering some significant physical properties are analyzed. We focus on the models where the cosmological observer measures isotropic radiation. We consider some examples, and a solution that models an ultrarelativistic gas is analyzed in detail., Comment: 16 pages, 7 figures

Stacking defects in noble metal nanoparticles significantly impact their optical, catalytic, and electrical properties. While some mechanisms behind their formation have been studied, the ability to deliberately manipulate nanoparticle bulk morphology remains largely unexplored. In this work, we introduce a pioneering mechanism - vacancy-driven twinning - that enables the transformation of face-centered cubic (fcc) gold into locally hexagonal close-packed (hcp) structures. This innovative approach, demonstrated through computational simulations, facilitates the creation of realistic , randomly multi-twinned nanoparticle models. By employing a recently developed multidomain X-ray diffraction method (MDXRD), we quantitatively assess the degree of twinning. It is a crucial step in transferring theoretical studies into practical applications. Our work aims to develop tools for modifying and controlling the bulk structure of fcc nanoparticles, Comment: 16 pages with 6 figures

Trusting software systems, particularly autonomous ones, is challenging. To address this, formal verification techniques can ensure these systems behave as expected. Runtime Verification (RV) is a leading, lightweight method for verifying system behaviour during execution. However, traditional RV assumes perfect information, meaning the monitoring component perceives everything accurately. This assumption often fails, especially with autonomous systems operating in real-world environments where sensors might be faulty. Additionally, traditional RV considers the monitor to be passive, lacking the capability to interpret the system's information and thus unable to address incomplete data. In this work, we extend standard RV of Linear Temporal Logic properties to accommodate scenarios where the monitor has imperfect information and behaves rationally. We outline the necessary engineering steps to update the verification pipeline and demonstrate our implementation in a case study involving robotic systems.

Finding proper collective variables for complex systems and processes is one of the most challenging tasks in simulations, which limits the interpretation of experimental and simulated data and the application of enhanced sampling techniques. Here, we propose a machine learning approach able to distill few, physically relevant variables by associating instantaneous configurations of the system to their corresponding inherent structures as defined in liquids theory. We apply this approach to the challenging case of structural transitions in nanoclusters, managing to characterize and explore the structural complexity of an experimentally relevant system constituted by 147 gold atoms. Our inherent-structure variables are shown to be effective at computing complex free-energy landscapes, transition rates, and at describing non-equilibrium melting and freezing processes. The effectiveness of this machine learning strategy guided by the generally-applicable concept of inherent structures shows promise to devise collective variables for a vast range of systems, including liquids, glasses, and proteins.

Multi-component aggregates are being intensively researched in various fields because of their highly tunable properties and wide applications. Due to the complex configurational space of these systems, research would greatly benefit from a general theoretical framework for the prediction of stable structures, which, however, is largely incomplete at present. Here we propose a general theory for the construction of multi-component icosahedral structures by assembling concentric shells of different chiral and achiral types, consisting of particles of different sizes. By mapping shell sequences into paths in the hexagonal lattice, we establish simple and general rules for designing a wide variety of magic icosahedral structures, and we evaluate the optimal size-mismatch between particles in the different shells. The predictions of our design strategy are confirmed by molecular dynamics simulations and density functional theory calculations for several multi-component atomic clusters and nanoparticles.

The azimuthal correlation angle, $\Delta\phi$, between the scattered lepton and the leading jet in deep inelastic $e^{\pm}p$ scattering at HERA has been studied using data collected with the ZEUS detector at a centre-of-mass energy of $\sqrt{s} = 318 \;\mathrm{GeV}$, corresponding to an integrated luminosity of $326 \;\mathrm{pb}^{-1}$. A measurement of jet cross sections in the laboratory frame was made in a fiducial region corresponding to photon virtuality $10 \;\mathrm{GeV}^2 < Q^2 < 350 \;\mathrm{GeV}^2$, inelasticity $0.04 < y < 0.7$, outgoing lepton energy $E_e > 10 \;\mathrm{GeV}$, lepton polar angle $140^\circ < \theta_e < 180^\circ$, jet transverse momentum $2.5 \;\mathrm{GeV} < p_\mathrm{T,jet} < 30 \;\mathrm{GeV}$, and jet pseudorapidity $-1.5 < \eta_\mathrm{jet} < 1.8$. Jets were reconstructed using the $k_\mathrm{T}$ algorithm with the radius parameter $R = 1$. The leading jet in an event is defined as the jet that carries the highest $p_\mathrm{T,jet}$. Differential cross sections, $d\sigma/d\Delta\phi$, were measured as a function of the azimuthal correlation angle in various ranges of leading-jet transverse momentum, photon virtuality and jet multiplicity. Perturbative calculations at $\mathcal{O}(\alpha_{s}^2)$ accuracy successfully describe the data within the fiducial region, although a lower level of agreement is observed near $\Delta\phi \rightarrow \pi$ for events with high jet multiplicity, due to limitations of the perturbative approach in describing soft phenomena in QCD. The data are equally well described by Monte Carlo predictions that supplement leading-order matrix elements with parton showering.

Sufficient statistic perturbation (SSP) is a widely used method for differentially private linear regression. SSP adopts a data-independent approach where privacy noise from a simple distribution is added to sufficient statistics. However, sufficient statistics can often be expressed as linear queries and better approximated by data-dependent mechanisms. In this paper we introduce data-dependent SSP for linear regression based on post-processing privately released marginals, and find that it outperforms state-of-the-art data-independent SSP. We extend this result to logistic regression by developing an approximate objective that can be expressed in terms of sufficient statistics, resulting in a novel and highly competitive SSP approach for logistic regression. We also make a connection to synthetic data for machine learning: for models with sufficient statistics, training on synthetic data corresponds to data-dependent SSP, with the overall utility determined by how well the mechanism answers these linear queries.

The current standard model of cosmology successfully describes a variety of measurements, but the nature of its main ingredients, dark matter and dark energy, remains unknown. Euclid is a medium-class mission in the Cosmic Vision 2015-2025 programme of the European Space Agency (ESA) that will provide high-resolution optical imaging, as well as near-infrared imaging and spectroscopy, over about 14,000 deg^2 of extragalactic sky. In addition to accurate weak lensing and clustering measurements that probe structure formation over half of the age of the Universe, its primary probes for cosmology, these exquisite data will enable a wide range of science. This paper provides a high-level overview of the mission, summarising the survey characteristics, the various data-processing steps, and data products. We also highlight the main science objectives and expected performance., Comment: Accepted for publication in the A&A special issue`Euclid on Sky'

While the third run of the Large Hadron Collider (LHC) is ongoing, the underlying theory that extends the Standard Model remains so far unknown. Left-Right Models (LRMs) introduce a new gauge sector, and can restore parity symmetry at high enough energies. If LRMs are indeed realized in nature, the mediators of the new weak force can be searched for in colliders via their direct production. We recast existing experimental limits from the LHC Run 2 and derive generic bounds on the masses of the heavy LRM gauge bosons. As a novelty, we discuss the dependence of the $W_R$ and $Z_R$ total decay width on the LRM scalar content, obtaining model-independent bounds within the specific realizations of the LRM scalar sectors analysed here. These bounds avoid the need to detail the spectrum of the scalar sector, and apply in the general case where no discrete symmetry is enforced. Moreover, we emphasize the impact on the $W_R$ production at LHC of general textures of the right-handed quark mixing matrix without manifest left-right symmetry. We find that the $W_R$ and $Z_R$ masses are constrained to lie above $2$ TeV and $4$ TeV, respectively., Comment: Contribution to the 2024 Electroweak session of the 58th Rencontres de Moriond

The rapid progress of research aimed at interpreting the inner workings of advanced language models has highlighted a need for contextualizing the insights gained from years of work in this area. This primer provides a concise technical introduction to the current techniques used to interpret the inner workings of Transformer-based language models, focusing on the generative decoder-only architecture. We conclude by presenting a comprehensive overview of the known internal mechanisms implemented by these models, uncovering connections across popular approaches and active research directions in this area.

Smart contracts are an attractive target for attackers, as evidenced by a long history of security incidents. A current limitation of smart contract verification tools is that they are not really effective in expressing and verifying liquidity properties regarding the exchange of crypto-assets: for example, is it true that in every reachable state a user can fire a sequence of transactions to withdraw a given amount of crypto-assets? We propose Solvent, a tool aimed at verifying these kinds of properties, which are beyond the reach of existing verification tools for Solidity. We evaluate the effectiveness and performance of Solvent through a common benchmark of smart contracts.

We present the LM Transparency Tool (LM-TT), an open-source interactive toolkit for analyzing the internal workings of Transformer-based language models. Differently from previously existing tools that focus on isolated parts of the decision-making process, our framework is designed to make the entire prediction process transparent, and allows tracing back model behavior from the top-layer representation to very fine-grained parts of the model. Specifically, it (1) shows the important part of the whole input-to-output information flow, (2) allows attributing any changes done by a model block to individual attention heads and feed-forward neurons, (3) allows interpreting the functions of those heads or neurons. A crucial part of this pipeline is showing the importance of specific model components at each step. As a result, we are able to look at the roles of model components only in cases where they are important for a prediction. Since knowing which components should be inspected is key for analyzing large models where the number of these components is extremely high, we believe our tool will greatly support the interpretability community both in research settings and in practical applications.

The growing interest in Large Language Models (LLMs) and in particular in conversational models with which users can interact has led to the development of a large number of open-source chat LLMs. These models are evaluated on a wide range of benchmarks to assess their capabilities in answering questions or solving problems on almost any possible topic or to test their ability to reason or interpret texts. Instead, the evaluation of the knowledge that these models have of the languages has received much less attention. For example, the words that they can recognize and use in different languages. In this paper, we evaluate the knowledge that open-source chat LLMs have of Spanish words by testing a sample of words in a reference dictionary. The results show that open-source chat LLMs produce incorrect meanings for an important fraction of the words and are not able to use most of the words correctly to write sentences with context. These results show how Spanish is left behind in the open-source LLM race and highlight the need to push for linguistic fairness in conversational LLMs ensuring that they provide similar performance across languages., Comment: Procesamiento del Lenguaje Natural, 73, 95-108

The verification of Multi-Agent Systems (MAS) poses a significant challenge. Various approaches and methodologies exist to address this challenge; however, tools that support them are not always readily available. Even when such tools are accessible, they tend to be hard-coded, lacking in compositionality, and challenging to use due to a steep learning curve. In this paper, we introduce a methodology designed for the formal verification of MAS in a modular and versatile manner, along with an initial prototype, that we named VITAMIN. Unlike existing verification methodologies and frameworks for MAS, VITAMIN is constructed for easy extension to accommodate various logics (for specifying the properties to verify) and models (for determining on what to verify such properties).

Information flows by routes inside the network via mechanisms implemented in the model. These routes can be represented as graphs where nodes correspond to token representations and edges to operations inside the network. We automatically build these graphs in a top-down manner, for each prediction leaving only the most important nodes and edges. In contrast to the existing workflows relying on activation patching, we do this through attribution: this allows us to efficiently uncover existing circuits with just a single forward pass. Additionally, the applicability of our method is far beyond patching: we do not need a human to carefully design prediction templates, and we can extract information flow routes for any prediction (not just the ones among the allowed templates). As a result, we can talk about model behavior in general, for specific types of predictions, or different domains. We experiment with Llama 2 and show that the role of some attention heads is overall important, e.g. previous token heads and subword merging heads. Next, we find similarities in Llama 2 behavior when handling tokens of the same part of speech. Finally, we show that some model components can be specialized on domains such as coding or multilingual texts.

The Young\'s modulus of a nylon string has been determined experimentally by combining elasticity theory and wave optics. A diffraction experiment has been setup to determine the change in the string diameter for different tensile forces applied by means of hanging weights. A linear elasticity model has been used to calculate the Young\'s modulus. A simple method has been provided which could be used as the basis of an experiment for introductory university level.

Supermartingales are here defined on a non-probabilistic setting and can be interpreted solely in terms of superhedging operations. The classical expectation operator is replaced by a pair of subadditive operators one of them providing a class of null sets and the other one acting as an outer integral. These operators are motivated by a financial theory of no-arbitrage pricing. Such a setting extends the classical stochastic framework by replacing the path space of the process by a trajectory set, while also providing a financial/gambling interpretation based on the notion of superhedging. The paper proves analogues of the following classical results: Doob's supermartingale decomposition and Doob's pointwise convergence theorem for nonnegative supermartingales. The approach shows how linearity of the expectation operator can be circumvented and how integrability properties in the proposed setting lead to the special case of (hedging) martingales while no integrability conditions are required for the general supermartingale case.

In this paper, we present the design and implementation of multi-trap optical tweezers based on new quadrifocal kinoform lenses. The phase distribution of these diffractive lenses is characterized by the Silver Mean sequence. The focusing properties of the resulting aperiodic DOEs coined Kinoform Silver Mean Lenses (KSMLs) are numerically examined. It is shown that, under monochromatic illumination, a KSML drives most of the incoming light into four single foci whose focal lengths are related to the Silver ratio. In this way, a KSML improves the diffraction efficiency of binary Fresnel Silver Mean Zone Plates. Through experimental results, the simultaneous trapping of particles in the four focal planes and their three-dimensional manipulation is demonstrated.

Multi-valued logics have a long tradition in the literature on system verification, including run-time verification. However, comparatively fewer model-checking tools have been developed for multi-valued specification languages. We present 3vLTL, a tool to generate Buchi automata from formulas in Linear-time Temporal Logic (LTL) interpreted on a three-valued semantics. Given an LTL formula, a set of atomic propositions as the alphabet for the automaton, and a truth value, our procedure generates a Buchi automaton that accepts all the words that assign the chosen truth value to the LTL formula. Given the particular type of the output of the tool, it can also be seamlessly processed by third-party libraries in a natural way. That is, the Buchi automaton can then be used in the context of formal verification to check whether an LTL formula is true, false, or undefined on a given model., Comment: In Proceedings FMAS 2023, arXiv:2311.08987

Faced with the complexities of managing natural gas-dependent power system amid the surge of renewable integration and load unpredictability, this study explores strategies for navigating emergency transitions to costlier secondary fuels. Our aim is to develop decision-support tools for operators during such exigencies. We approach the problem through a Markov Decision Process (MDP) framework, accounting for multiple uncertainties. These include the potential for dual-fuel generator failures and operator response during high-pressure situations. Additionally, we consider the finite reserves of primary fuel, governed by gas-flow partial differential equations (PDEs) and constrained by nodal pressure. Other factors include the variability in power forecasts due to renewable generation and the economic impact of compulsory load shedding. For tractability, we address the MDP in a simplified context, replacing it by Markov Processes evaluated against a selection of policies and scenarios for comparison. Our study considers two models for the natural gas system: an over-simplified model tracking linepack and a more nuanced model that accounts for gas flow network heterogeneity. The efficacy of our methods is demonstrated using a realistic model replicating Israel's power-gas infrastructure., Comment: Accepted for presentation at the IEEE Conference on Decision and Control (CDC), 2024

The Checkerboard conformal field theory is an interesting representative of a large class of non-unitary, logarithmic Fishnet CFTs (FCFT) in arbitrary dimension which have been intensively studied in the last years. Its planar Feynman graphs have the structure of a regular square lattice with checkerboard colouring. Such graphs are integrable since each coloured cell of the lattice is equal to an R-matrix in the principal series representations of the conformal group. We compute perturbatively and numerically the anomalous dimension of the shortest single-trace operator in two reductions of the Checkerboard CFT: the first one corresponds to the Fishnet limit of the twisted ABJM theory in 3D, whereas the spectrum in the second, 2D reduction contains the energy of the BFKL Pomeron. We derive an analytic expression for the Checkerboard analogues of Basso--Dixon 4-point functions, as well as for the class of Diamond-type 4-point graphs with disc topology. The properties of the latter are studied in terms of OPE for operators with open indices. We prove that the spectrum of the theory receives corrections only at even orders in the loop expansion and we conjecture such a modification of Checkerboard CFT where quantum corrections occur only with a given periodicity in the loop order., Comment: 63 pages, 24 figures, v2: typos fixed, references added, prepared or submission to JHEP

In many power systems, particularly those isolated from larger intercontinental grids, operational dependence on natural gas becomes pivotal, especially during fluctuations or unavailability of renewables coupled with uncertain consumption patterns. Efficient orchestration and inventive strategies are imperative for the smooth functioning of these standalone gas-grid systems. This paper delves into the challenge of synchronized dynamic and stochastic optimization for independent transmission-level gas-grid systems. Our approach's novelty lies in amalgamating the staggered-grid method for the direct assimilation of gas-flow PDEs with an automated sensitivity analysis facilitated by SciML/Julia, further enhanced by an intuitive linkage between gas and power grids via nodal flows. We initiate with a single pipe to establish a versatile and expandable methodology, later showcasing its effectiveness with increasingly intricate examples., Comment: 7 pages, 7 figures, submitted to PSCC 2024

The model checking problem for multi-agent systems against Strategy Logic specifications is known to be non-elementary. On this logic several fragments have been defined to tackle this issue but at the expense of expressiveness. In this paper, we propose a three-valued semantics for Strategy Logic upon which we define an abstraction method. We show that the latter semantics is an approximation of the classic two-valued one for Strategy Logic. Furthermore, we extend MCMAS, an open-source model checker for multi-agent specifications, to incorporate our abstraction method and present some promising experimental results.

We construct a general class of modified Ellis wormholes, where one asymptotic Minkowski region is replaced by a bounded 2-sphere core, characterized by asymptotic finite areal radius. We pursue an in-depth analysis of the resulting geometry, outlining that geodesic completeness is guaranteed also when the radial function asymptotically shrinks to zero. Then, we study the evolution of scalar perturbations, bringing out how these geometric configurations can in principle affect the time-domain profiles of quasinormal modes, pointing out the distinctive features with respect to other black holes or wormholes geometries.

Autonomous and robotic systems are increasingly being trusted with sensitive activities with potentially serious consequences if that trust is broken. Runtime verification techniques present a natural source of inspiration for monitoring and enforcing the desirable properties of the communication protocols in place, providing a formal basis and ways to limit intrusiveness. A recently proposed approach, RV-TEE, shows how runtime verification can enhance the level of trust to the Rich Execution Environment (REE), consequently adding a further layer of protection around the Trusted Execution Environment (TEE). By reflecting on the implication of deploying RV in the context of trustworthy computing, we propose practical solutions to two threat models for the RV-TEE monitoring process: one where the adversary has gained access to the system without elevated privileges, and another where the adversary gains all privileges to the host system but fails to steal secrets from the TEE., Comment: In Proceedings AREA 2023, arXiv:2310.00333

The volume comprises the proceedings of the Third Workshop on Agents and Robots for reliable Engineered Autonomy (AREA 2023), held alongside the 26th European Conference on Artificial Intelligence (ECAI 2023). It explores the convergence of autonomous agents and robotics, emphasizing the practical application of agents in real-world scenarios with physical interactions. The workshop highlights the growing importance of enhanced autonomy and reliable behavior in robotic systems and the need for novel verification and validation methods. Its primary objective is to promote collaboration between researchers in these fields, aiming to address complex challenges in autonomous robotic systems. The volume includes 7 full papers and 5 short papers.

We investigate the time evolution of spherically symmetric boson stars in Palatini $f(\mathcal{R})$ gravity through Numerical Relativity computations. Employing a novel approach that establishes a correspondence between modified gravity with scalar matter and General Relativity with modified scalar matter, we are able to use the techniques of Numerical Relativity to simulate these systems. Specifically, we focus on the quadratic theory $f(\mathcal{R})=\mathcal{R}+\xi\mathcal{R}^2$ and compare the obtained solutions with those in General Relativity, exploring both positive and negative values of the coupling parameter $\xi$. Our findings reveal that boson stars in Palatini $f(\mathcal{R})$ gravity exhibit both stable and unstable evolutions. The latter give rise to three distinct scenarios: migration towards a stable configuration, complete dispersion, and gravitational collapse leading to the formation of a baby universe structure.

The necessary and sufficient conditions for a spacetime with an invariant frame to admit a group of isometries of dimension $r$ are given in terms of the connection tensor $H$ associated with this frame. In Petrov-Bel types I, II and III, and in other spacetimes where an invariant frame algebraically defined by the curvature tensor exists, the connection tensor $H$ is given in terms of the Weyl and Ricci tensors without an explicit determination of the frame. Thus, an IDEAL (intrinsic, deductive, explicit and algorithmic) characterization of these spacetimes follows. Some examples show that this algorithm can be easily implemented on the xAct Mathematica suite of packages., Comment: 19 pages, 1 figure

While the third run of the Large Hadron Collider (LHC) is ongoing, the underlying theory that extends the Standard Model remains so far unknown. Left-Right Models (LRMs) introduce a new gauge sector, and can restore parity symmetry at high enough energies. If LRMs are indeed realized in nature, the mediators of the new weak force can be searched for in colliders via their direct production. We recast existing experimental bounds from LHC Run 2 on the heavy LRM gauge boson masses. As a novelty, we discuss the effect of the LRM scalar content on the total width of the new gauge bosons, obtaining model-independent bounds within the specific realizations of the LRM scalar sectors analysed here. These bounds avoid the need to detail the spectrum of the scalar sector, and apply in the general case where no discrete symmetry is enforced. Moreover, we emphasize the effect of the structure of the quark right-handed mixing matrix on the charged LRM gauge boson production at LHC. We find that $W_R$ and $Z_R$ masses are constrained to lie above $2$ TeV and $4$ TeV, respectively.

We analyze a family of large language models in such a lightweight manner that can be done on a single GPU. Specifically, we focus on the OPT family of models ranging from 125m to 66b parameters and rely only on whether an FFN neuron is activated or not. First, we find that the early part of the network is sparse and represents many discrete features. Here, many neurons (more than 70% in some layers of the 66b model) are "dead", i.e. they never activate on a large collection of diverse data. At the same time, many of the alive neurons are reserved for discrete features and act as token and n-gram detectors. Interestingly, their corresponding FFN updates not only promote next token candidates as could be expected, but also explicitly focus on removing the information about triggering them tokens, i.e., current input. To the best of our knowledge, this is the first example of mechanisms specialized at removing (rather than adding) information from the residual stream. With scale, models become more sparse in a sense that they have more dead neurons and token detectors. Finally, some neurons are positional: them being activated or not depends largely (or solely) on position and less so (or not at all) on textual data. We find that smaller models have sets of neurons acting as position range indicators while larger models operate in a less explicit manner.

A new measurement of inclusive-jet cross sections in the Breit frame in neutral current deep inelastic scattering using the ZEUS detector at the HERA collider is presented. The data were taken in the years 2004 to 2007 at a centre-of-mass energy of $318\,\text{GeV}$ and correspond to an integrated luminosity of $347\,\text{pb}^{-1}$. Massless jets, reconstructed using the $k_t$-algorithm in the Breit reference frame, have been measured as a function of the squared momentum transfer, $Q^2$, and the transverse momentum of the jets in the Breit frame, $p_{\perp,\text{Breit}}$. The measured jet cross sections are compared to previous measurements and to perturbative QCD predictions. The measurement has been used in a next-to-next-to-leading-order QCD analysis to perform a simultaneous determination of parton distribution functions of the proton and the strong coupling, resulting in a value of $\alpha_s(M_Z^2) = 0.1142 \pm 0.0017~\text{(experimental/fit)}$ ${}^{+0.0006}_{-0.0007}~\text{(model/parameterisation)}$ ${}^{+0.0006}_{-0.0004}~\text{(scale)}$, whose accuracy is improved compared to similar measurements. In addition, the running of the strong coupling is demonstrated using data obtained at different scales., Comment: 42 pages, 10 figures

Behavioral testing in NLP allows fine-grained evaluation of systems by examining their linguistic capabilities through the analysis of input-output behavior. Unfortunately, existing work on behavioral testing in Machine Translation (MT) is currently restricted to largely handcrafted tests covering a limited range of capabilities and languages. To address this limitation, we propose to use Large Language Models (LLMs) to generate a diverse set of source sentences tailored to test the behavior of MT models in a range of situations. We can then verify whether the MT model exhibits the expected behavior through matching candidate sets that are also generated using LLMs. Our approach aims to make behavioral testing of MT systems practical while requiring only minimal human effort. In our experiments, we apply our proposed evaluation framework to assess multiple available MT systems, revealing that while in general pass-rates follow the trends observable from traditional accuracy-based metrics, our method was able to uncover several important differences and potential bugs that go unnoticed when relying only on accuracy.