Search

Super-resolution techniques have the potential to reduce the computational cost of cosmological and astrophysical simulations. This can be achieved by enabling traditional simulation methods to run at lower resolution and then efficiently computing high-resolution data corresponding to the simulated low-resolution data. In this work, we investigate the application of a Wasserstein Generative Adversarial Network (WGAN) to increase the particle resolution of dark-matter-only simulations, reproducing and building on prior results. Our WGAN models successfully generate high-resolution data with summary statistics, including the power spectrum and halo mass function, that closely match those of true high-resolution simulations. We also identify a limitation of the WGAN model in the form of smeared features in the generated high-resolution snapshots, particularly in the shapes of dark-matter halos.

As Large Language Models (LLMs) and generative AI become increasingly widespread, concerns about content safety have grown in parallel. Currently, there is a clear lack of high-quality, human-annotated datasets that address the full spectrum of LLM-related safety risks and are usable for commercial applications. To bridge this gap, we propose a comprehensive and adaptable taxonomy for categorizing safety risks, structured into 12 top-level hazard categories with an extension to 9 fine-grained subcategories. This taxonomy is designed to meet the diverse requirements of downstream users, offering more granular and flexible tools for managing various risk types. Using a hybrid data generation pipeline that combines human annotations with a multi-LLM "jury" system to assess the safety of responses, we obtain Aegis 2.0, a carefully curated collection of 34,248 samples of human-LLM interactions, annotated according to our proposed taxonomy. To validate its effectiveness, we demonstrate that several lightweight models, trained using parameter-efficient techniques on Aegis 2.0, achieve performance competitive with leading safety models fully fine-tuned on much larger, non-commercial datasets. In addition, we introduce a novel training blend that combines safety with topic following data.This approach enhances the adaptability of guard models, enabling them to generalize to new risk categories defined during inference. We plan to open-source Aegis 2.0 data and models to the research community to aid in the safety guardrailing of LLMs., Comment: arXiv admin note: text overlap with arXiv:2404.05993

We generate Gaussian radial function based higher order compact RBF-FD formulas for some differential operators. Analytical expressions for weights associated to first and second derivative formulas (up to order 10) and 2D-Laplacian formulas (up to order 6) are derived. Then these weights are used to obtain analytical expression for local truncation errors. The weights are obtained by symbolic computation of a linear system in Mathematica. Often such linear systems are not directly amenable to symbolic computation. We make use of symmetry of formula stencil along with Taylor series expansions for performing the computation. In the flat limit, the formulas converge to their respective order polynomial based compact FD formulas. We validate the formulas with standard test functions and demonstrate improvement in approximation accuracy with respect to corresponding order multiquadric based compact RBF-FD formulas and compact FD schemes. We also compute optimal value of shape parameter for each formula., Comment: 18 pages

The growing adoption of machine learning (ML) in modelling atmospheric and oceanic processes offers a promising alternative to traditional numerical methods. It is essential to benchmark the performance of both ML and physics-informed ML (PINN) models to evaluate their predictive skill, particularly for short- to medium-term forecasting. In this study, we utilize gridded sea surface temperature (SST) data and six atmospheric predictors (cloud cover, relative humidity, solar radiation, surface pressure, u-component of velocity, and v-component of velocity) to capture both spatial and temporal patterns in SST predictions.

Recent advancements in large language models have significantly expedited the process of generating front-end code. This allows users to rapidly prototype user interfaces and ideate through code, a process known as exploratory programming. However, existing LLM code generation tools focus more on technical implementation details rather than finding the right design given a particular problem. We present DynEx, an LLM-based method for design exploration in accelerated exploratory programming. DynEx introduces a technique to explore the design space through a structured Design Matrix before creating the prototype with a modular, stepwise approach to LLM code generation. Code is generated sequentially, and users can test and approve each step before moving onto the next. A user study of 10 experts found that DynEx increased design exploration and enabled the creation of more complex and varied prototypes compared to a Claude Artifact baseline. We conclude with a discussion of the implications of design exploration for exploratory programming., Comment: 27 pages, 9 figures

Advancing Machine Learning (ML)-based perception models for autonomous systems necessitates addressing weak spots within the models, particularly in challenging Operational Design Domains (ODDs). These are environmental operating conditions of an autonomous vehicle which can contain difficult conditions, e.g., lens flare at night or objects reflected in a wet street. This report introduces a novel methodology for training with augmentations to enhance model robustness and performance in such conditions. The proposed approach leverages customized physics-based augmentation functions, to generate realistic training data that simulates diverse ODD scenarios. We present a comprehensive framework that includes identifying weak spots in ML models, selecting suitable augmentations, and devising effective training strategies. The methodology integrates hyperparameter optimization and latent space optimization to fine-tune augmentation parameters, ensuring they maximally improve the ML models' performance. Experimental results demonstrate improvements in model performance, as measured by commonly used metrics such as mean Average Precision (mAP) and mean Intersection over Union (mIoU) on open-source object detection and semantic segmentation models and datasets. Our findings emphasize that optimal training strategies are model- and data-specific and highlight the benefits of integrating augmentations into the training pipeline. By incorporating augmentations, we observe enhanced robustness of ML-based perception models, making them more resilient to edge cases encountered in real-world ODDs. This work underlines the importance of customized augmentations and offers an effective solution for improving the safety and reliability of autonomous driving functions.

Large language models (LLMs) are remarkably good at writing code. A particularly valuable case of human-LLM collaboration is code-based UI prototyping, a method for creating interactive prototypes that allows users to view and fully engage with a user interface. We conduct a formative study of GPT Pilot, a leading LLM-generated code-prototyping system, and find that its inflexibility towards change once development has started leads to weaknesses in failure prevention and dynamic planning; it closely resembles the linear workflow of the waterfall model. We introduce DIDUP, a system for code-based UI prototyping that follows an iterative spiral model, which takes changes and iterations that come up during the development process into account. We propose three novel mechanisms for LLM-generated code-prototyping systems: (1) adaptive planning, where plans should be dynamic and reflect changes during implementation, (2) code injection, where the system should write a minimal amount of code and inject it instead of rewriting code so users have a better mental model of the code evolution, and (3) lightweight state management, a simplified version of source control so users can quickly revert to different working states. Together, this enables users to rapidly develop and iterate on prototypes., Comment: 5 pages, 3 figures

Large language models (LLMs) have recently shown tremendous promise in serving as the backbone to agentic systems, as demonstrated by their performance in multi-faceted, challenging benchmarks like SWE-Bench and Agent-Bench. However, to realize the true potential of LLMs as autonomous agents, they must learn to identify, call, and interact with external tools and application program interfaces (APIs) to complete complex tasks. These tasks together are termed function calling. Endowing LLMs with function calling abilities leads to a myriad of advantages, such as access to current and domain-specific information in databases and knowledge sources, and the ability to outsource tasks that can be reliably performed by tools, e.g., a Python interpreter or calculator. While there has been significant progress in function calling with LLMs, there is still a dearth of open models that perform on par with proprietary LLMs like GPT, Claude, and Gemini. Therefore, in this work, we introduce the GRANITE-20B-FUNCTIONCALLING model under an Apache 2.0 license. The model is trained using a multi-task training approach on seven fundamental tasks encompassed in function calling, those being Nested Function Calling, Function Chaining, Parallel Functions, Function Name Detection, Parameter-Value Pair Detection, Next-Best Function, and Response Generation. We present a comprehensive evaluation on multiple out-of-domain datasets comparing GRANITE-20B-FUNCTIONCALLING to more than 15 other best proprietary and open models. GRANITE-20B-FUNCTIONCALLING provides the best performance among all open models on the Berkeley Function Calling Leaderboard and fourth overall. As a result of the diverse tasks and datasets used for training our model, we show that GRANITE-20B-FUNCTIONCALLING has better generalizability on multiple tasks in seven different evaluation datasets.

Dialogue policies play a crucial role in developing task-oriented dialogue systems, yet their development and maintenance are challenging and typically require substantial effort from experts in dialogue modeling. While in many situations, large amounts of conversational data are available for the task at hand, people lack an effective solution able to extract dialogue policies from this data. In this paper, we address this gap by first illustrating how Large Language Models (LLMs) can be instrumental in extracting dialogue policies from datasets, through the conversion of conversations into a unified intermediate representation consisting of canonical forms. We then propose a novel method for generating dialogue policies utilizing a controllable and interpretable graph-based methodology. By combining canonical forms across conversations into a flow network, we find that running graph traversal algorithms helps in extracting dialogue flows. These flows are a better representation of the underlying interactions than flows extracted by prompting LLMs. Our technique focuses on giving conversation designers greater control, offering a productivity tool to improve the process of developing dialogue policies.

We release the Nemotron-4 340B model family, including Nemotron-4-340B-Base, Nemotron-4-340B-Instruct, and Nemotron-4-340B-Reward. Our models are open access under the NVIDIA Open Model License Agreement, a permissive model license that allows distribution, modification, and use of the models and its outputs. These models perform competitively to open access models on a wide range of evaluation benchmarks, and were sized to fit on a single DGX H100 with 8 GPUs when deployed in FP8 precision. We believe that the community can benefit from these models in various research studies and commercial applications, especially for generating synthetic data to train smaller language models. Notably, over 98% of data used in our model alignment process is synthetically generated, showcasing the effectiveness of these models in generating synthetic data. To further support open research and facilitate model development, we are also open-sourcing the synthetic data generation pipeline used in our model alignment process.

We present an approach to build Large Language Model (LLM) based slot-filling system to perform Dialogue State Tracking in conversational assistants serving across a wide variety of industry-grade applications. Key requirements of this system include: 1) usage of smaller-sized models to meet low latency requirements and to enable convenient and cost-effective cloud and customer premise deployments, and 2) zero-shot capabilities to serve across a wide variety of domains, slot types and conversational scenarios. We adopt a fine-tuning approach where a pre-trained LLM is fine-tuned into a slot-filling model using task specific data. The fine-tuning data is prepared carefully to cover a wide variety of slot-filling task scenarios that the model is expected to face across various domains. We give details of the data preparation and model building process. We also give a detailed analysis of the results of our experimental evaluations. Results show that our prescribed approach for slot-filling model building has resulted in 6.9% relative improvement of F1 metric over the best baseline on a realistic benchmark, while at the same time reducing the latency by 57%. More over, the data we prepared has helped improve F1 on an average by 4.2% relative across various slot-types.

High-quality preference datasets are essential for training reward models that can effectively guide large language models (LLMs) in generating high-quality responses aligned with human preferences. As LLMs become stronger and better aligned, permissively licensed preference datasets, such as Open Assistant, HH-RLHF, and HelpSteer need to be updated to remain effective for reward modeling. Methods that distil preference data from proprietary LLMs such as GPT-4 have restrictions on commercial usage imposed by model providers. To improve upon both generated responses and attribute labeling quality, we release HelpSteer2, a permissively licensed preference dataset (CC-BY-4.0). Using a powerful internal base model trained on HelpSteer2, we are able to achieve the SOTA score (92.0%) on Reward-Bench's primary dataset, outperforming currently listed open and proprietary models, as of June 12th, 2024. Notably, HelpSteer2 consists of only ten thousand response pairs, an order of magnitude fewer than existing preference datasets (e.g., HH-RLHF), which makes it highly efficient for training reward models. Our extensive experiments demonstrate that reward models trained with HelpSteer2 are effective in aligning LLMs. In particular, we propose SteerLM 2.0, a model alignment approach that can effectively make use of the rich multi-attribute score predicted by our reward models. HelpSteer2 is available at https://huggingface.co/datasets/nvidia/HelpSteer2 and code is available at https://github.com/NVIDIA/NeMo-Aligner

The celebrated Aharonov-Bohm effect is perhaps the first example in which the the interplay between classical topology and quantum theory was explored. This connection has continued to shed light on diverse areas of physics like quantum statistics, anomalies, condensed matter physics, and gauge theories. Several attempts were made to generalize the Aharonov-Bohm effect by modifying the simple solenoidal current distribution used by them to the case of multiple solenoids, and a toroidal solenoid, for example. A particularly ambitious task is to confine the magnetic flux to the interior of a knotted solenoid. While it is to be expected that a non-trivial phase factor will be picked up by the wave function of a charged particle travelling in the complement of the knot in three-dimensional space, the lack of symmetry defied attempts to explicitly solve the associated scattering problem. In this paper we report on a way to make progress towards this problem based on multipole expansions. The vector potential produced by a knot is obtained by making a multipole expansion, which is then used to calculate the S-matrix for the scattering of the charged particle, in the Born approximation. It is found that the S-matrix carries an imprint of the knottedness at the octopole order. For the case of a torus knot, a curious factorization property is seen to hold., Comment: 26 pages

The time evolution of an open quantum system is governed by the Gorini-Kossakowski-Sudarshan-Lindlad equation for the reduced density operator of the system. This operator is obtained from the full density operator of the composite system involving the system itself, the bath, and the interactions between them, by performing a partial trace over the bath degrees of freedom. The entanglement between the system and the bath leads to a generalized Liouville evolution that involves, amongst other things, dissipation and decoherence of the system. In a similar fashion, the time evolution of a physical observable in a classically constrained dynamical system is governed by a generalization of the Liouville equation, in which the usual Poisson bracket is replaced by the so-called Dirac bracket. The generalization takes into account the reduction in the phase space of the system because of constraints which arise either because they are introduced by hand, or because of singular Lagrangians. We derive an intriguing, but precise classical-quantum correspondence between the aforementioned situations which connects the Lindblad operators to the constraints. The correspondence is illustrated in the ubiquitous example of coupled simple harmonic oscillators studied earlier in various contexts like quantum optics, dissipative quantum systems, Brownian motion, and the area law for the entropy of black holes., Comment: v3: 28 pages, relevant details of the Constrained system and Lindblad equation have been moved to the appendix from the review section

Recent advancements in instruction-tuning datasets have predominantly focused on specific tasks like mathematical or logical reasoning. There has been a notable gap in data designed for aligning language models to maintain topic relevance in conversations - a critical aspect for deploying chatbots to production. We introduce the CantTalkAboutThis dataset to help language models remain focused on the subject at hand during task-oriented interactions. It consists of synthetic dialogues on a wide range of conversation topics from different domains. These dialogues are interspersed with distractor turns that intentionally divert the chatbot from the predefined topic. Fine-tuning language models on this dataset helps make them resilient to deviating from the role assigned and improves their ability to maintain topical coherence compared to general-purpose instruction-tuned LLMs like GPT-4-turbo and Mixtral-Instruct. Additionally, preliminary observations suggest that training models on this dataset also enhance their performance on fine-grained instruction following tasks, including safety alignment.

We propose a variational autoencoder (VAE)-based model for building forward and inverse structure-property linkages, a problem of paramount importance in computational materials science. Our model systematically combines VAE with regression, linking the two models through a two-level prior conditioned on the regression variables. The regression loss is optimized jointly with the reconstruction loss of the variational autoencoder, learning microstructure features relevant for property prediction and reconstruction. The resultant model can be used for both forward and inverse prediction i.e., for predicting the properties of a given microstructure as well as for predicting the microstructure required to obtain given properties. Since the inverse problem is ill-posed (one-to-many), we derive the objective function using a multi-modal Gaussian mixture prior enabling the model to infer multiple microstructures for a target set of properties. We show that for forward prediction, our model is as accurate as state-of-the-art forward-only models. Additionally, our method enables direct inverse inference. We show that the microstructures inferred using our model achieve desired properties reasonably accurately, avoiding the need for expensive optimization loops., Comment: 12 pages main paper, 9 pages appendix. 10 tables and 11 figures. Accepted for publication in PAKDD 2024

Periodically driven thermodynamic systems support stable non-equilibrium oscillating states with properties drastically different from equilibrium. They exhibit even more exotic features for low viscous drives, which is a regime that is hard to probe due to singular behavior of the underlying Langevin dynamics near vanishing viscosity. We propose a method, based on singular perturbation and Floquet theories, that allows us to obtain oscillating states in this limit. We then find two distinct classes of distributions, each exhibiting interesting features that can be exploited for a range of practical applicability, including cooling a system and triggering chemical reactions through weakly interacting driven environments., Comment: 6+13 pages, 2+4 figures

When creating policies, plans, or designs for people, it is challenging for designers to foresee all of the ways in which people may reason and behave. Recently, Large Language Models (LLMs) have been shown to be able to simulate human reasoning. We extend this work by measuring LLMs ability to simulate strategic reasoning in the ultimatum game, a classic economics bargaining experiment. Experimental evidence shows human strategic reasoning is complex; people will often choose to punish other players to enforce social norms even at personal expense. We test if LLMs can replicate this behavior in simulation, comparing two structures: single LLMs and multi-agent systems. We compare their abilities to (1) simulate human-like reasoning in the ultimatum game, (2) simulate two player personalities, greedy and fair, and (3) create robust strategies that are logically complete and consistent with personality. Our evaluation shows that multi-agent systems are more accurate than single LLMs (88 percent vs. 50 percent) in simulating human reasoning and actions for personality pairs. Thus, there is potential to use LLMs to simulate human strategic reasoning to help decision and policy-makers perform preliminary explorations of how people behave in systems., Comment: 10 pages, 6 figures

We implemented Population III (Pop. III) star formation in mini-halos within the MERAXES semi-analytic galaxy formation and reionisation model, run on top of a N-body simulation with $L = 10 h^{-1}$ cMpc with 2048$^3$ particles resolving all dark matter halos down to the mini-halos ($\sim 10^5 M_\odot$). Our modelling includes the chemical evolution of the IGM, with metals released through supernova-driven bubbles that expand according to the Sedov-Taylor model. We found that SN-driven metal bubbles are generally small, with radii typically of 150 ckpc at z = 6. Hence, the majority of the first galaxies are likely enriched by their own star formation. However, as reionization progresses, the feedback effects from the UV background become more pronounced, leading to a halt in star formation in low-mass galaxies, after which external chemical enrichment becomes more relevant. We explore the sensitivity of the star formation rate density and stellar mass functions on the unknown values of free parameters. We also discuss the observability of Pop. III dominated systems with JWST, finding that the inclusion of Pop. III galaxies can have a significant effect on the total UV luminosity function at z = 12 - 16. Our results support the idea that the excess of bright galaxies detected with JWST might be explained by the presence of bright top-heavy Pop. III dominated galaxies without requiring an increased star formation efficiency., Comment: 19 Pages, 22 Figures, 2 tables, 2 Appendices - Accepted for publication in MNRAS (14th of Feb. 2024) - Meraxes is now publicly available!

LaNiGa$_2$ is superconductor that breaks time-reversal symmetry in the superconducting state without any known nearby magnetism. Recently, single crystals of LaNiGa$_2$ have been synthesized, revealing a nonsymmorphic Cmcm space group. Here, we report measurements of the electronic structure of LaNiGa$_2$ throughout the three-dimensional Brillouin zone (BZ) using angle-resolved photoemission spectroscopy (ARPES). Our findings show broad consistency with density functional theory (DFT) calculations and provide evidence for degeneracies in the electronic structure that are predicted from the space group. The calculations also predict four Fermi surfaces which cross the purported nodal plane and should therefore form two degenerate pairs. We report evidence for those predicted symmetry enforced degeneracies as well as accidental near degeneracies throughout the BZ. These degeneracies and near-degeneracies may play a role in the pairing mechanism of LaNiGa$_2$. Our results provide insight into the interplay between structure, Fermiology, and superconductivity in unconventional superconductors with nonsymmorphic space group., Comment: Main: 8 pages, 6 figures. SI: 6 pages, 8 figures

Compressing large language models (LLMs), often consisting of billions of parameters, provides faster inference, smaller memory footprints, and enables local deployment. Two standard compression techniques are pruning and quantization, with the former eliminating redundant connections in model layers and the latter representing model parameters with fewer bits. The key tradeoff is between the degree of compression and the impact on the quality of the compressed model. Existing research on LLM compression primarily focuses on performance in terms of general metrics like perplexity or downstream task accuracy. More fine-grained metrics, such as those measuring parametric knowledge, remain significantly underexplored. To help bridge this gap, we present a comprehensive analysis across multiple model families (ENCODER, ENCODER-DECODER, and DECODER) using the LAMA and LM-HARNESS benchmarks in order to systematically quantify the effect of commonly employed compression techniques on model performance. A particular focus is on tradeoffs involving parametric knowledge, with the goal of providing practitioners with practical insights to help make informed decisions on compression. We release our codebase1 to enable further research., Comment: Accepted to EMNLP 2023 Findings