Retrieval-augmented generation

Retrieval pipelines, vector search, chunking and reranking: how models are grounded in a corpus instead of their weights.

6,299 stories · RSS feed

arXiv Machine Learning

Sep 23

MIND the Gap: A Geographic Implicit Neural Representation with Adjustable Spatial Scale

The paper introduces MIND, a method that distills specialist geospatial model embeddings into a single generalist coordinate embedding with adjustable spatial granularity, using nested supervision across multiple embedding dimensions. MIND’s design allows downstream predictors to use only leading chunks or apply a Chunked Penalty to downweight finer details without retraining the INR. The authors evaluate MIND on CoordBench, a large INR benchmark of 52 datasets and 78 targets, and report that MIND and its Chunked Penalty variant achieve the highest regression and classification scores, especially under regional holdout, establishing a new state‑of‑the‑art for geographic implicit neural representations.

By Isaac Corley, Arjun Rao, Esther Rolf, Konstantin Klemmer, Evan Shelhamer, Nils Lehmann, Marc Ru{\ss}wurm, Gengchen Mai, Nathan Jacobs, Hannah Kerner

rag efficiency benchmarks

arXiv Machine Learning

Sep 23

Optimizing Denoising Trajectories in dLLMs: A Lightweight Evolutionary Heuristic Approach

The paper introduces a lightweight evolutionary heuristic scheduler for Diffusion Large Language Models (dLLMs) that optimizes denoising trajectories using CMA-ES. It addresses failure modes of existing confidence-based schedulers—EOS Overflow and Proximal Bias—by integrating multiple heuristic features with a contextual mean-field embedding, requiring only 393 trainable parameters. Evaluations on LLaDA and Dream across reasoning and planning benchmarks show consistent outperformance over strong baselines and recent state‑of‑the‑art methods.

By Zijian Zhao, Dian Jin, Xialiang Tong, Sen Li, Mingxuan Yuan

llms rag diffusion benchmarks safety

arXiv Machine Learning

Sep 23

Foundation model embeddings capture pre-diagnostic changes on screening mammograms

The study examined whether embeddings from four foundation models—Mammo-CLIP, HOPPR, MedImageInsight, and BiomedCLIP—could detect pre‑diagnostic changes in screening mammograms. Using 1,773 biopsied women and matched controls, the researchers measured the speed of movement along a data‑derived “cancer direction” in embedding space over successive screening intervals. They found that embeddings from clinically grounded models (Mammo‑CLIP, HOPPR, MedImageInsight) showed faster drift in malignant cases compared to controls, while the general biomedical model BiomedCLIP did not, indicating that foundation model embeddings can encode early tissue changes without task‑specific fine‑tuning.

By Kalina P. Slavkova, Eric Brattain, Aditya Gowd, Akash Pattnaik, Jean-Benoit Delbrouck, Matthew Morgan, Julie Bauml, Javid Abderezaei, Khan Siddiqui

rag multimodal

arXiv Machine Learning

Sep 23

Transport-Coupled Bayesian Flows for Molecular Graph Generation

Transport-Coupled Bayesian Flows for Molecular Graph Generation (TopBF) addresses a key mismatch in existing diffusion models for molecular graph generation by eliminating the need for hard discretization during sampling. The framework generates graphs directly in continuous parameter distributions, learns graph topology via a Quasi-Wasserstein optimal‑transport coupling with geodesic costs, and enables property‑conditioned generation without retraining. Experiments on QM9 and ZINC250k show that TopBF achieves higher structural fidelity and more efficient generation compared to prior methods.

By Yida Xiong, Jiameng Chen, Kun Li, Hongzhi Zhang, Xiantao Cai, Lei Lei, Wenbin Hu

rag diffusion