Running Large Language Models (LLMs) on edge devices is crucial for reducing latency, improving real-time processing, and enhancing privacy. By performing inference directly on the device, data does not need to be sent to the cloud, ensuring faster responses and reducing reliance on network connectivity. However, implementing LLMs on edge devices presents challenges, particularly with managing key-value (KV) caches, which plays a pivotal role in LLM serving. As the input text lengthens, the size of the KV cache increases linearly with the sequence length, leading to a significant memory footprint and data access costs. On the other hand, edge devices have limited memory and computational power, making it hard to store and efficiently access the large caches needed for LLM inference. To mitigate the substantial overhead caused by KV cache, we propose using embedded DRAM (eDRAM) as the primary storage for LLM serving in edge device, which offers higher storage density compared to SRAM. However, to ensure data integrity, eDRAM needs periodic refresh operations, which are power-intensive. To reduce eDRAM costs and improve overall system performance, we propose~\textit{Kelle}, a software-hardware co-design solution optimized for deploying LLMs on eDRAM-based edge systems. Combined with our fine-grained memory eviction, recomputation, and refresh control algorithms, the \textit{Kelle} accelerator delivers a $3.9\times$ speedup and $4.5\times$ energy savings compared to existing baseline solutions.

As wildfires become increasingly destructive and expensive to control, effective management of active wildfires requires accurate, real-time fire spread predictions. To enhance the forecasting accuracy of active fires, data assimilation plays a vital role by integrating observations (such as remote-sensing data) and fire predictions generated from numerical models. This paper provides a comprehensive investigation on the application of a recently proposed diffusion-model-based filtering algorithm -- the Ensemble Score Filter (EnSF) -- to the data assimilation problem for real-time active wildfire spread predictions. Leveraging a score-based generative diffusion model, EnSF has been shown to have superior accuracy for high-dimensional nonlinear filtering problems, making it an ideal candidate for the filtering problems of wildfire spread models. Technical details are provided, and our numerical investigations demonstrate that EnSF provides superior accuracy, stability, and computational efficiency, establishing it as a robust and practical method for wildfire data assimilation. Our code has been made publicly available.

Time series data are often affected by various forms of corruption, such as missing values, noise, and outliers, which pose significant challenges for tasks such as forecasting and anomaly detection. To address these issues, inverse problems focus on reconstructing the original signal from corrupted data by leveraging prior knowledge about its underlying structure. While deep learning methods have demonstrated potential in this domain, they often require extensive pretraining and struggle to generalize under distribution shifts. In this work, we propose RINS-T (Robust Implicit Neural Solvers for Time Series Linear Inverse Problems), a novel deep prior framework that achieves high recovery performance without requiring pretraining data. RINS-T leverages neural networks as implicit priors and integrates robust optimization techniques, making it resilient to outliers while relaxing the reliance on Gaussian noise assumptions. To further improve optimization stability and robustness, we introduce three key innovations: guided input initialization, input perturbation, and convex output combination techniques. Each of these contributions strengthens the framework's optimization stability and robustness. These advancements make RINS-T a flexible and effective solution for addressing complex real-world time series challenges. Our code is available at https://github.com/EPFL-IMOS/RINS-T.

When we post to a group chat or talk to an AI chatbot, we don't think about how these technologies came to be. We take it for granted we can instantly communicate. We only notice the importance and reach of these systems when they're not accessible. Companies describe these systems with metaphors such as the "cloud" or "artificial intelligence", suggesting something intangible. But they are deeply material.

Anthropic Has a Plan to Keep Its AI From Building a Nuclear Weapon. Anthropic partnered with the US government to create a filter meant to block Claude from helping someone build a nuke. Experts are divided on whether its a necessary protection--or a protection at all. At the end of August, the AI company Anthropic announced that its chatbot Claude wouldn't help anyone build a nuclear weapon. According to Anthropic, it had partnered with the Department of Energy (DOE) and the National Nuclear Security Administration (NNSA) to make sure Claude wouldn't spill nuclear secrets.

Latent-variable energy-based models (LV-EBMs) assign a single normalized energy to joint pairs of observed data and latent variables, offering expressive generative modeling while capturing hidden structure. We recast maximum-likelihood training as a saddle problem over distributions on the latent and joint manifolds and view the inner updates as coupled Wasserstein gradient flows. The resulting algorithm alternates overdamped Langevin updates for a joint negative pool and for conditional latent particles with stochastic parameter ascent, requiring no discriminator or auxiliary networks. We prove existence and convergence under standard smoothness and dissi-pativity assumptions, with decay rates in KL divergence and Wasserstein-2 distance. The saddle-point view further yields an ELBO strictly tighter than bounds obtained with restricted amortized posteriors. Our method is evaluated on numerical approximations of physical systems and performs competitively against comparable approaches.

Calibration sample selection and forecast combination are two simple yet powerful tools used in forecasting. They can be combined with a variety of models to significantly improve prediction accuracy, at the same time offering easy implementation and low computational complexity. While their effectiveness has been repeatedly confirmed in prior scientific literature, the topic is still underexplored in the field of electricity price forecasting. In this research article we apply the Autoregressive Hybrid Nearest Neighbors (ARHNN) method to three long-term time series describing the German, Spanish and New England electricity markets. We show that it outperforms popular literature benchmarks in terms of forecast accuracy by up to 10%. We also propose two simplified variants of the method, granting a vast decrease in computation time with only minor loss of prediction accuracy. Finally, we compare the forecasts' performance in a battery storage system trading case study. We find that using a forecast-driven strategy can achieve up to 80% of theoretical maximum profits while trading, demonstrating business value in practical applications.

Machine-learned force fields (MLFFs), especially pre-trained foundation models, are transforming computational materials science by enabling ab initio-level accuracy at molecular dynamics scales. Yet their rapid rise raises a key question: should researchers train specialist models from scratch, fine-tune generalist foundation models, or use hybrid approaches? The trade-offs in data efficiency, accuracy, cost, and robustness to out-of-distribution failure remain unclear. We introduce a benchmarking framework using defect migration pathways, evaluated through nudged elastic band trajectories, as diagnostic probes that test both interpolation and extrapolation. Using Cr-doped Sb2Te3 as a representative two-dimensional material, we benchmark multiple training paradigms within the MACE architecture across equilibrium, kinetic (atomic migration), and mechanical (interlayer sliding) tasks. Fine-tuned models substantially outperform from-scratch and zero-shot approaches for kinetic properties but show partial loss of long-range physics. Representational analysis reveals distinct, non-overlapping latent encodings, indicating that different training strategies learn different aspects of system physics. This framework provides practical guidelines for MLFF development and establishes migration-based probes as efficient diagnostics linking performance to learned representations, guiding future uncertainty-aware active learning.

The development of accurate machine learning interatomic potentials (MLIPs) is limited by the fragmented availability and inconsistent formatting of quantum mechanical trajectory datasets derived from Density Functional Theory (DFT). These datasets are expensive to generate yet difficult to combine due to variations in format, metadata, and accessibility. To address this, we introduce LeMat-Traj, a curated dataset comprising over 120 million atomic configurations aggregated from large-scale repositories, including the Materials Project, Alexandria, and OQMD. LeMat-Traj standardizes data representation, harmonizes results and filters for high-quality configurations across widely used DFT functionals (PBE, PBESol, SCAN, r2SCAN). It significantly lowers the barrier for training transferrable and accurate MLIPs. LeMat-Traj spans both relaxed low-energy states and high-energy, high-force structures, complementing molecular dynamics and active learning datasets. By fine-tuning models pre-trained on high-force data with LeMat-Traj, we achieve a significant reduction in force prediction errors on relaxation tasks. We also present LeMaterial-Fetcher, a modular and extensible open-source library developed for this work, designed to provide a reproducible framework for the community to easily incorporate new data sources and ensure the continued evolution of large-scale materials datasets. LeMat-Traj and LeMaterial-Fetcher are publicly available at https://huggingface.co/datasets/LeMaterial/LeMat-Traj and https://github.com/LeMaterial/lematerial-fetcher.

To address the challenges posed by the heavy reliance of multi-output models on preset probability distributions and embedded prior knowledge in non-injective regression tasks, this paper proposes a cycle consistency-based data-driven training framework. The method jointly optimizes a forward model Φ: X to Y and a backward model Ψ: Y to X, where the cycle consistency loss is defined as L _cycleb equal L(Y reduce Φ(Ψ(Y))) (and vice versa). By minimizing this loss, the framework establishes a closed-loop mechanism integrating generation and validation phases, eliminating the need for manual rule design or prior distribution assumptions. Experiments on normalized synthetic and simulated datasets demonstrate that the proposed method achieves a cycle reconstruction error below 0.003, achieving an improvement of approximately 30% in evaluation metrics compared to baseline models without cycle consistency. Furthermore, the framework supports unsupervised learning and significantly reduces reliance on manual intervention, demonstrating potential advantages in non-injective regression tasks.