Welcome to Gist.Science
We read the latest papers from arXiv, bioRxiv, and medRxiv and produce easy-to-understand explanations, key takeaways, and technical summaries — in ten languages.
Imagine the world of Large Language Models (LLMs) as a massive, bustling city. The people building these models are like city planners trying to make sure the city is safe from criminals (hackers).…
Too many papers, too little time. Get daily digests of the most novel papers matching your exact research interests — with technical summaries in your language.
XFP is a dynamic, calibration-free weight quantization method for LLM inference that automatically determines codebook parameters based on user-specified quality thresholds, effectively separating sparse outliers to achieve high throughput and accuracy while enabling massive models to fit within constrained memory via its quality-driven "H-Process" iteration.
This paper introduces a coordinate-free Riemannian Nyström approximation for efficiently constructing low-rank tangent operators on manifolds using Haar–Grassmann sketching, which enables a faster randomized Newton-type optimization method while preserving positive semidefiniteness and accuracy.
This paper proposes and validates language-independent statistical features derived from co-occurrence and run-length matrices for string similarity computation and classification, demonstrating their superior performance over existing state-of-the-art measures in both synthetic experiments and real-world plagiarism detection tasks.
GPart introduces a highly parameter-efficient fine-tuning method that achieves end-to-end isometry by mapping a single low-dimensional trainable vector directly into the full weight space via a global partition matrix, thereby eliminating the distance-distorting low-rank bottleneck of LoRA while delivering state-of-the-art performance across diverse tasks.
This paper answers a question posed by Simental by offering combinatorial interpretations of a formula generalizing rational Catalan numbers, which arises in the study of Springer fibers, through the lenses of binary necklaces and anchored Dyck paths.
The paper introduces PARADIS, a physics-inspired neural architecture that improves weather forecasting by explicitly decomposing the prediction process into distinct advection, diffusion, and reaction blocks, utilizing a differentiable Neural Semi-Lagrangian operator to efficiently model long-range transport while achieving competitive skill and superior spectral fidelity on ERA5 benchmarks.
This paper identifies and quantifies "cultural anachronism"—the tendency of Vision-Language Models to misinterpret historical artifacts using temporally inappropriate concepts—by introducing the TAB-VLM benchmark, which reveals significant performance gaps in state-of-the-art models when reasoning about Indian cultural heritage.
This paper presents a simple abstract model demonstrating that regionalizing e-retail fulfillment networks, contrary to the traditional belief in global economies of scale, can significantly reduce order fulfillment delays by characterizing equilibrium assignments under a greedy strategy and providing algorithmic solutions for low-delay configurations.
This paper demonstrates how decomposing automorphism groups by analyzing twins and co-twins simplifies the study of symmetry parameters and automorphism structures in vertex-transitive graphs, with a specific focus on circulant graphs.
The paper introduces Orchard, an open-source framework featuring a lightweight environment layer (Orchard Env) and specialized training recipes that enable scalable, high-performance agentic modeling across coding, GUI, and personal assistant domains, achieving state-of-the-art results among open-source models.
This paper introduces Conservative Peng's Q() (CPQL), a model-free offline multi-step reinforcement learning algorithm that leverages a conservative Peng's Q() operator to achieve implicit behavior regularization and near-optimal performance, significantly outperforming existing single-step baselines on D4RL benchmarks while also enhancing offline-to-online learning frameworks.