Search

Mott metal-insulator transitions possess electronic, magnetic, and structural degrees of freedom promising next generation energy-efficient electronics. We report a previously unknown, hierarchically ordered state during a Mott transition and demonstrate correlated switching of functional electronic properties. We elucidate in-situ formation of an intrinsic supercrystal in a Ca2RuO4 thin film. Machine learning-assisted X-ray nanodiffraction together with electron microscopy reveal multi-scale periodic domain formation at and below the film transition temperature (TFilm ~ 200-250 K) and a separate anisotropic spatial structure at and above TFilm. Local resistivity measurements imply an intrinsic coupling of the supercrystal orientation to the material's anisotropic conductivity. Our findings add an additional degree of complexity to the physical understanding of Mott transitions, opening opportunities for designing materials with tunable electronic properties.

Instruction-based image editing holds immense potential for a variety of applications, as it enables users to perform any editing operation using a natural language instruction. However, current models in this domain often struggle with accurately executing user instructions. We present Emu Edit, a multi-task image editing model which sets state-of-the-art results in instruction-based image editing. To develop Emu Edit we train it to multi-task across an unprecedented range of tasks, such as region-based editing, free-form editing, and Computer Vision tasks, all of which are formulated as generative tasks. Additionally, to enhance Emu Edit's multi-task learning abilities, we provide it with learned task embeddings which guide the generation process towards the correct edit type. Both these elements are essential for Emu Edit's outstanding performance. Furthermore, we show that Emu Edit can generalize to new tasks, such as image inpainting, super-resolution, and compositions of editing tasks, with just a few labeled examples. This capability offers a significant advantage in scenarios where high-quality samples are scarce. Lastly, to facilitate a more rigorous and informed assessment of instructable image editing models, we release a new challenging and versatile benchmark that includes seven different image editing tasks.

Hadwiger's conjecture in convex geometry, formulated in 1957, states that every convex body in $\mathbb{R}^n$ can be covered by $2^n$ translations of its interior. Despite significant efforts, the best known bound related to this problem was $\mathcal{O}(4^n \sqrt{n} \log n)$ for more than sixty years. In 2021, Huang, Slomka, Tkocz, and Vritsiou made a major breakthrough by improving the estimate by a factor of $\exp\left(\Omega(\sqrt{n})\right)$. Further, for $\psi_2$ bodies they proved that at most $\exp(-\Omega(n))\cdot4^n$ translations of its interior are needed to cover it. Through a probabilistic approach we show that the bound $\exp(-\Omega(n))\cdot4^n$ can be obtained for convex bodies with sufficiently many well-behaved sub-gaussian marginals. Using a small diameter approximation, we present how the currently best known bound for the general case, due to Campos, Van Hintum, Morris, and Tiba can also be deduced from our results., Comment: We found a discrepancy between our definition of the $\psi_2$ constant and the referenced article. As a result Theorem 1.1 does not generalize the previously known bound as the constant should be $b_2 L_K$

Large Language Models (LLMs) represent a revolution in AI. However, they also pose many significant risks, such as the presence of biased, private, copyrighted or harmful text. For this reason we need open, transparent and safe solutions. We introduce a complete open-source ecosystem for developing and testing LLMs. The goal of this project is to boost open alternatives to closed-source approaches. We release h2oGPT, a family of fine-tuned LLMs of diverse sizes. We also introduce H2O LLM Studio, a framework and no-code GUI designed for efficient fine-tuning, evaluation, and deployment of LLMs using the most recent state-of-the-art techniques. Our code and models are fully open-source. We believe this work helps to boost AI development and make it more accessible, efficient and trustworthy. The demo is available at: https://gpt.h2o.ai/, Comment: EMNLP 2023 Demo - ACL Empirical Methods in Natural Language Processing

Traditional theories of the NMR autocorrelation function for intramolecular dipole pairs assume single-exponential decay, yet the calculated autocorrelation of realistic systems display a rich, multi-exponential behavior resulting in anomalous NMR relaxation dispersion (i.e., frequency dependence). We develop an approach to model and interpret the multi-exponential autocorrelation using simple, physical models within a rigorous statistical mechanical development that encompasses both rotational and translational diffusion in the same framework. We recast the problem of evaluating the autocorrelation in terms of averaging over a diffusion propagator whose evolution is described by a Fokker-Planck equation. The time-independent part admits an eigenfunction expansion, allowing us to write the propagator as a sum over modes. Each mode has a spatial part that depends on the specified eigenfunction, and a temporal part that depends on the corresponding eigenvalue (i.e., correlation time) with a simple, exponential decay. The spatial part is a probability distribution of the dipole-pair, analogous to the stationary states of a quantum harmonic oscillator. Drawing inspiration from the idea of inherent structures in liquids, we interpret each of the spatial contributions as a specific molecular mode. These modes can be used to model and predict NMR dipole-dipole relaxation dispersion of fluids by incorporating phenomena on the molecular level. We validate our statistical mechanical description of the distribution in molecular modes with molecular dynamics simulations interpreted without any relaxation models or adjustable parameters: the most important poles in the Pad{\'e}-Laplace transform of the simulated autocorrelation agree with the eigenvalues predicted by the theory.

We address the challenge of making spatial audio datasets by proposing a shared mechanized recording space that can run custom acoustic experiments: a Mechatronic Acoustic Research System (MARS). To accommodate a wide variety of experiments, we implement an extensible architecture for wireless multi-robot coordination which enables synchronized robot motion for dynamic scenes with moving speakers and microphones. Using a virtual control interface, we can remotely design automated experiments to collect large-scale audio data. This data is shown to be similar across repeated runs, demonstrating the reliability of MARS. We discuss the potential for MARS to make audio data collection accessible for researchers without dedicated acoustic research spaces., Comment: 5 pages, 5 figures, IWAENC 2022

This manuscript is a research resource description and presents a large and novel Electronic Health Records (EHR) data resource, American Family Cohort (AFC). The AFC data is derived from Centers for Medicare and Medicaid Services (CMS) certified American Board of Family Medicine (ABFM) PRIME registry. The PRIME registry is the largest national Qualified Clinical Data Registry (QCDR) for Primary Care. The data is converted to a popular common data model, the Observational Health Data Sciences and Informatics (OHDSI) Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). The resource presents approximately 90 million encounters for 7.5 million patients. All 100% of the patients present age, gender, and address information, and 73% report race. Nealy 93% of patients have lab data in LOINC, 86% have medication data in RxNorm, 93% have diagnosis in SNOWMED and ICD, 81% have procedures in HCPCS or CPT, and 61% have insurance information. The richness, breadth, and diversity of this research accessible and research ready data is expected to accelerate observational studies in many diverse areas. We expect this resource to facilitate research in many years to come.

This paper introduces a novel graph-based filter method for automatic feature selection (abbreviated as GB-AFS) for multi-class classification tasks. The method determines the minimum combination of features required to sustain prediction performance while maintaining complementary discriminating abilities between different classes. It does not require any user-defined parameters such as the number of features to select. The methodology employs the Jeffries-Matusita (JM) distance in conjunction with t-distributed Stochastic Neighbor Embedding (t-SNE) to generate a low-dimensional space reflecting how effectively each feature can differentiate between each pair of classes. The minimum number of features is selected using our newly developed Mean Simplified Silhouette (abbreviated as MSS) index, designed to evaluate the clustering results for the feature selection task. Experimental results on public data sets demonstrate the superior performance of the proposed GB-AFS over other filter-based techniques and automatic feature selection approaches. Moreover, the proposed algorithm maintained the accuracy achieved when utilizing all features, while using only $7\%$ to $30\%$ of the features. Consequently, this resulted in a reduction of the time needed for classifications, from $15\%$ to $70\%$., Comment: 8 pages, 4 figures

We present CM3Leon (pronounced "Chameleon"), a retrieval-augmented, token-based, decoder-only multi-modal language model capable of generating and infilling both text and images. CM3Leon uses the CM3 multi-modal architecture but additionally shows the extreme benefits of scaling up and tuning on more diverse instruction-style data. It is the first multi-modal model trained with a recipe adapted from text-only language models, including a large-scale retrieval-augmented pre-training stage and a second multi-task supervised fine-tuning (SFT) stage. It is also a general-purpose model that can do both text-to-image and image-to-text generation, allowing us to introduce self-contained contrastive decoding methods that produce high-quality outputs. Extensive experiments demonstrate that this recipe is highly effective for multi-modal models. CM3Leon achieves state-of-the-art performance in text-to-image generation with 5x less training compute than comparable methods (zero-shot MS-COCO FID of 4.88). After SFT, CM3Leon can also demonstrate unprecedented levels of controllability in tasks ranging from language-guided image editing to image-controlled generation and segmentation.