Using Aberrations to Improve Dose-Efficient Tilt-corrected 4D-STEM Imaging

Kavli Affiliate: David A. Muller | First 5 Authors: Desheng Ma, Desheng Ma, , , | Summary: Tilt-corrected imaging methods in four-dimensional scanning transmission electron microscopy (4D-STEM) have recently emerged as a new class of direct ptychography methods that are especially useful at low dose. The operation of tilt correction unfolds the contrast transfer functions […]


Continue.. Using Aberrations to Improve Dose-Efficient Tilt-corrected 4D-STEM Imaging

Simulating Spectral Confusion in SPHEREx Photometry and Redshifts

Kavli Affiliate: James J. Bock | First 5 Authors: Zhaoyu Huai, Zhaoyu Huai, , , | Summary: We model the impact of source confusion on photometry and the resulting spectrophotometric redshifts for SPHEREx, a NASA Medium-Class Explorer that is carrying out an all-sky near-infrared spectral survey. Spectral confusion from untargeted background galaxies degrades sensitivity and […]


Continue.. Simulating Spectral Confusion in SPHEREx Photometry and Redshifts

Understanding the Mixture-of-Experts with Nadaraya-Watson Kernel

Kavli Affiliate: Ting Xu | First 5 Authors: Chuanyang Zheng, Chuanyang Zheng, , , | Summary: Mixture-of-Experts (MoE) has become a cornerstone in recent state-of-the-art large language models (LLMs). Traditionally, MoE relies on $mathrmSoftmax$ as the router score function to aggregate expert output, a designed choice that has persisted from the earliest MoE models to […]


Continue.. Understanding the Mixture-of-Experts with Nadaraya-Watson Kernel

Understanding the Mixture-of-Experts with Nadaraya-Watson Kernel

Kavli Affiliate: Ting Xu | First 5 Authors: Chuanyang Zheng, Chuanyang Zheng, , , | Summary: Mixture-of-Experts (MoE) has become a cornerstone in recent state-of-the-art large language models (LLMs). Traditionally, MoE relies on $mathrmSoftmax$ as the router score function to aggregate expert output, a designed choice that has persisted from the earliest MoE models to […]


Continue.. Understanding the Mixture-of-Experts with Nadaraya-Watson Kernel

Translation from Wearable PPG to 12-Lead ECG

Kavli Affiliate: Wei Gao | First 5 Authors: Hui Ji, Hui Ji, , , | Summary: The 12-lead electrocardiogram (ECG) is the gold standard for cardiovascular monitoring, offering superior diagnostic granularity and specificity compared to photoplethysmography (PPG). However, existing 12-lead ECG systems rely on cumbersome multi-electrode setups, limiting sustained monitoring in ambulatory settings, while current […]


Continue.. Translation from Wearable PPG to 12-Lead ECG

jina-reranker-v3: Last but Not Late Interaction for Listwise Document Reranking

Kavli Affiliate: Feng Wang | First 5 Authors: Feng Wang, Feng Wang, , , | Summary: jina-reranker-v3 is a 0.6B-parameter multilingual listwise reranker that introduces a novel "last but not late" interaction. Unlike late interaction models like ColBERT that encode documents separately before multi-vector matching, our approach applies causal attention between the query and all […]


Continue.. jina-reranker-v3: Last but Not Late Interaction for Listwise Document Reranking

AMS-L0 silicon ladder assembly with high precision by gantry

Kavli Affiliate: Feng Wang | First 5 Authors: Feng Wang, Feng Wang, , , | Summary: A high-precision silicon microstrip detector assembly methodology based on a gantry system is presented in this paper. The proposed approach has been applied to the mass production of all flight-model detector modules for the L0 tracking detector in the […]


Continue.. AMS-L0 silicon ladder assembly with high precision by gantry

Experience Paper: Adopting Activity Recognition in On-demand Food Delivery Business

Kavli Affiliate: Wei Gao | First 5 Authors: , , , , | Summary: This paper presents the first nationwide deployment of human activity recognition (HAR) technology in the on-demand food delivery industry. We successfully adapted the state-of-the-art LIMU-BERT foundation model to the delivery platform. Spanning three phases over two years, the deployment progresses from […]


Continue.. Experience Paper: Adopting Activity Recognition in On-demand Food Delivery Business

AdaPtis: Reducing Pipeline Bubbles with Adaptive Pipeline Parallelism on Heterogeneous Models

Kavli Affiliate: Wei Gao | First 5 Authors: Jihu Guo, Jihu Guo, , , | Summary: Pipeline parallelism is widely used to train large language models (LLMs). However, increasing heterogeneity in model architectures exacerbates pipeline bubbles, thereby reducing training efficiency. Existing approaches overlook the co-optimization of model partition, model placement, and workload scheduling, resulting in […]


Continue.. AdaPtis: Reducing Pipeline Bubbles with Adaptive Pipeline Parallelism on Heterogeneous Models