Recent breakthroughs in solving reasoning, math and coding problems with Large Language Models (LLMs) have been enabled by investing substantial computation budgets at inference time. Therefore, inference speed is one of the most critical properties of LLM architectures, and there is a growing need for LLMs that are efficient and fast at inference. Recently, LLMs built on the xLSTM architecture have emerged as a powerful alternative to Transformers, offering linear compute scaling with sequence length and constant memory usage, both highly desirable properties for efficient inference. However, such xLSTM-based LLMs have yet to be scaled to larger models and assessed and compared with respect to inference speed and efficiency. In this work, we introduce xLSTM 7B, a 7-billion-parameter LLM that combines xLSTM's architectural benefits with targeted optimizations for fast and efficient inference. Our experiments demonstrate that xLSTM 7B achieves performance on downstream tasks comparable to other similar-sized LLMs, while providing significantly faster inference speeds and greater efficiency compared to Llama- and Mamba-based LLMs. These results establish xLSTM 7B as the fastest and most efficient 7B LLM, offering a solution for tasks that require large amounts of test-time computation. Our work highlights xLSTM's potential as a foundational architecture for methods building on heavy use of LLM inference. Our model weights, model code and training code are open-source.

The rapid evolution of Artificial Intelligence (AI) and Machine Learning (ML) has significantly transformed logistics and supply chain management, particularly in the pursuit of sustainability and eco-efficiency. This study explores AI-based methodologies for optimizing logistics operations in the USA, focusing on reducing environmental impact, improving fuel efficiency, and minimizing costs. Key AI applications include predictive analytics for demand forecasting, route optimization through machine learning, and AI-powered fuel efficiency strategies. Various models, such as Linear Regression, XGBoost, Support Vector Machine, and Neural Networks, are applied to real-world logistics datasets to reduce carbon emissions based on logistics operations, optimize travel routes to minimize distance and travel time, and predict future deliveries to plan optimal routes. Other models such as K-Means and DBSCAN are also used to optimize travel routes to minimize distance and travel time for logistics operations. This study utilizes datasets from logistics companies' databases. The study also assesses model performance using metrics such as mean absolute error (MAE), mean squared error (MSE), and R2 score. This study also explores how these models can be deployed to various platforms for real-time logistics and supply chain use. The models are also examined through a thorough case study, highlighting best practices and regulatory frameworks that promote sustainability. The findings demonstrate AI's potential to enhance logistics efficiency, reduce carbon footprints, and contribute to a more resilient and adaptive supply chain ecosystem.

The transition of the electricity production from fossil fuels to renewable sources is one of the most important tasks of the present and future. State financed feed-in tariffs have helped to kick-off this transition. Today though, the production from renewable sources is more and more brought to liberalised electricity markets using so-called Green Power Purchase Agreements (PPA). PPAs are contracts that sell the production of a generating asset at a fixed price. PPAs are not new, they have helped to secure the large investment sums needed when building conventional power plants for decades. This is achieved by guaranteeing a secure stream of income to the owner of the asset. At the same time the buyer is exposed to market risk. A PPA is called "Green" when its underlying generating asset uses a renewable source. Due to the fact that electricity is not storable its production has to match consumption at every point in time.

Knowledge driven discovery of novel materials necessitates the development of the causal models for the property emergence. While in classical physical paradigm the causal relationships are deduced based on the physical principles or via experiment, rapid accumulation of observational data necessitates learning causal relationships between dissimilar aspects of materials structure and functionalities based on observations. For this, it is essential to integrate experimental data with prior domain knowledge. Here we demonstrate this approach by combining high-resolution scanning transmission electron microscopy (STEM) data with insights derived from large language models (LLMs). By fine-tuning ChatGPT on domain-specific literature, such as arXiv papers on ferroelectrics, and combining obtained information with data-driven causal discovery, we construct adjacency matrices for Directed Acyclic Graphs (DAGs) that map the causal relationships between structural, chemical, and polarization degrees of freedom in Sm-doped BiFeO3 (SmBFO). This approach enables us to hypothesize how synthesis conditions influence material properties, particularly the coercive field (E0), and guides experimental validation. The ultimate objective of this work is to develop a unified framework that integrates LLM-driven literature analysis with data-driven discovery, facilitating the precise engineering of ferroelectric materials by establishing clear connections between synthesis conditions and their resulting material properties.

Time series frequently manifest distribution shifts, diverse latent features, and non-stationary learning dynamics, particularly in open and evolving environments. These characteristics pose significant challenges for out-of-distribution (OOD) generalization. While substantial progress has been made, a systematic synthesis of advancements remains lacking. To address this gap, we present the first comprehensive review of OOD generalization methodologies for time series, organized to delineate the field's evolutionary trajectory and contemporary research landscape. We organize our analysis across three foundational dimensions: data distribution, representation learning, and OOD evaluation. For each dimension, we present several popular algorithms in detail. Furthermore, we highlight key application scenarios, emphasizing their real-world impact. Finally, we identify persistent challenges and propose future research directions. A detailed summary of the methods reviewed for the generalization of OOD in time series can be accessed at https://tsood-generalization.com.

Automatic classification of electric power quality events with respect to their root causes is critical for electrical grid management. In this paper, we present comparative evaluation results of an extensive set of machine learning models for the classification of power quality events, based on their root causes. After extensive experiments using different machine learning libraries, it is observed that the best performing learning models turn out to be Cubic SVM and XGBoost. During error analysis, it is observed that the main source of performance degradation for both models is the classification of ABC faults as ABCG faults, or vice versa. Ultimately, the models achieving the best results will be integrated into the event classification module of a large-scale power quality and grid monitoring system for the Turkish electricity transmission system.

Geospatial raster data, such as that collected by satellite-based imaging systems at different times and spectral bands, hold immense potential for enabling a wide range of high-impact applications. This potential stems from the rich information that is spatially and temporally contextualized across multiple channels and sensing modalities. Recent work has adapted existing self-supervised learning approaches for such geospatial data. However, they fall short of scalable model architectures, leading to inflexibility and computational inefficiencies when faced with an increasing number of channels and modalities. To address these limitations, we introduce Low-rank Efficient Spatial-Spectral Vision Transformer with three key innovations: i) the LESS Attention Block that approximates high-dimensional spatial-spectral attention through Kronecker's product of the low-dimensional spatial and spectral attention components; ii) the Continuous Positional-Channel Embedding Layer that preserves both the continuity and physical characteristics of each spatial-spectral patch; and iii) the Perception Field Mask that exploits local spatial dependencies by constraining attention to neighboring patches. To evaluate the proposed innovations, we construct GFM-Bench, which serves as a comprehensive benchmark for such geospatial raster data. We pretrain LESS ViT using a Hyperspectral Masked Autoencoder framework with integrated positional and channel masking strategies. Experimental results demonstrate that our proposed method achieves competitive performance against state-of-the-art multi-modal geospatial foundation models while outperforming them on cross-satellite generalization tasks with higher computational efficiency. The flexibility and extensibility of our framework make it a promising direction for future geospatial data analysis tasks that involve a wide range of modalities and channels.

Foundation models have had a significant impact across various AI applications, enabling use cases that were previously impossible. Contrastive Visual Language Models (VLMs), in particular, have outperformed other techniques in many tasks. However, their prevalence in remote sensing (RS) is still limited, due to the scarcity of diverse remote-sensing visual-language datasets. In this work we introduce two novel image-caption datasets for training of remote sensing foundation models. The first dataset pairs aerial and satellite imagery with captions generated by Gemini using landmarks extracted from Google Maps. The second dataset utilizes public web images and their corresponding alt-text, filtered for the remote sensing domain, resulting in a diverse dataset with greater breadth in image styles and subject matter. These datasets are used to pre-train the MaMMUT~\citep{kuo2023mammutsimplearchitecturejoint} VLM architecture, resulting in state-of-the-art generalization performance in zero-shot cross-modal retrieval on well-known public benchmarks. Finally, we present our ongoing research to distill image-level knowledge gained in the VLM contrastive training procedure to enhance the model's localization ability. Specifically, we iteratively generate pseudo-labels for image regions based on the model's attention maps and use these labels for further training. To mitigate noisy attention maps and create robust segmentation masks, we introduce a novel attention-pooling mechanism called the Smooth-Attention-Operation.

The use of large language models (LLMs) to simulate human behavior has gained significant attention, particularly through personas that approximate individual characteristics. Persona-based simulations hold promise for transforming disciplines that rely on population-level feedback, including social science, economic analysis, marketing research, and business operations. Traditional methods to collect realistic persona data face significant challenges. They are prohibitively expensive and logistically challenging due to privacy constraints, and often fail to capture multi-dimensional attributes, particularly subjective qualities. Consequently, synthetic persona generation with LLMs offers a scalable, cost-effective alternative. However, current approaches rely on ad hoc and heuristic generation techniques that do not guarantee methodological rigor or simulation precision, resulting in systematic biases in downstream tasks. Through extensive large-scale experiments including presidential election forecasts and general opinion surveys of the U.S. population, we reveal that these biases can lead to significant deviations from real-world outcomes. Our findings underscore the need to develop a rigorous science of persona generation and outline the methodological innovations, organizational and institutional support, and empirical foundations required to enhance the reliability and scalability of LLM-driven persona simulations. To support further research and development in this area, we have open-sourced approximately one million generated personas, available for public access and analysis at https://huggingface.co/datasets/Tianyi-Lab/Personas.

Action-constrained reinforcement learning (ACRL) is a generic framework for learning control policies with zero action constraint violation, which is required by various safety-critical and resource-constrained applications. The existing ACRL methods can typically achieve favorable constraint satisfaction but at the cost of either high computational burden incurred by the quadratic programs (QP) or increased architectural complexity due to the use of sophisticated generative models. In this paper, we propose a generic and computationally efficient framework that can adapt a standard unconstrained RL method to ACRL through two modifications: (i) To enforce the action constraints, we leverage the classic acceptance-rejection method, where we treat the unconstrained policy as the proposal distribution and derive a modified policy with feasible actions. (ii) To improve the acceptance rate of the proposal distribution, we construct an augmented two-objective Markov decision process (MDP), which include additional self-loop state transitions and a penalty signal for the rejected actions. This augmented MDP incentives the learned policy to stay close to the feasible action sets. Through extensive experiments in both robot control and resource allocation domains, we demonstrate that the proposed framework enjoys faster training progress, better constraint satisfaction, and a lower action inference time simultaneously than the state-of-the-art ACRL methods. We have made the source code publicly available to encourage further research in this direction.