NVIDIA/Megatron-LM
Ongoing research training transformer models at scale (⭐ 15534)
Ongoing research training transformer models at scale (⭐ 15534)
Build–operate–transfer (BOT) or build–own–operate–transfer (BOOT) is a form of project delivery method, usually for large-scale infrastructure projects
Many computers and devices are becoming more connected to the internet in recent years; the use of the Internet Protocol (IP) has made the connectivity and identification of these devices possible in large scale. In this paper, we will discuss the ev...
Detecting fashion landmarks is a fundamental technique for visual clothing analysis. Due to the large variation and non-rigid deformation of clothes, localizing fashion landmarks suffers from large spatial variances across poses, scales, and styles....
Multi-agent predictive modeling is an essential step for understanding physical, social and team-play systems. Recently, Interaction Networks (INs) were proposed for the task of modeling multi-agent physical systems, INs scale with the number of inte...
radiometric dating, the technique can be used to construct and calibrate the geomagnetic polarity time scale. This is one of the dating methodologies
Screening mechanisms allow light scalar fields to dynamically avoid the constraints that come from our lack of observation of a long-range fifth force. Galactic scale tests are of particular interest when the light scalar is introduced to explain the...
Finding a set of nested partitions of a dataset is useful to uncover relevant structure at different scales, and is often dealt with a data-dependent methodology. In this paper, we introduce a general two-step methodology for model-based hierarchical...
Interactive programming with interleaved code snippet cells and natural language markdown is recently gaining popularity in the form of Jupyter notebooks, which accelerate prototyping and collaboration. To study code generation conditioned on a long...
Large-scale analysis of pedestrian infrastructures, particularly sidewalks, is critical to human-centric urban planning and design. Benefiting from the rich data set of planimetric features and high-resolution orthoimages provided through the New Yor...
We examine the potential of using large-scale open crowdsourced sidewalk data from Project Sidewalk to study the distribution and condition of sidewalks in Seattle, WA. While potentially noisier than professionally gathered sidewalk datasets, crowdso...
Dataset distillation synthesizes a small set of images from a large-scale real dataset such that synthetic and real images share similar behavioral properties (e.g, distributions of gradients or features) during a training process. Through extensive...
The article describes observations of the crescent of Venus and other bright planets using a large camera obscura. The goal of the article is to demonstrate that these observations can be successfully performed. Achieving positive results depends on...
Artificial Intelligence (AI) is increasingly becoming a trusted advisor in people's lives. A new concern arises if AI persuades people to break ethical rules for profit. Employing a large-scale behavioural experiment (N = 1,572), we test whether AI-g...
Construction of a nuclear weapon requires access to kilogram-scale quantities of fissile material, which can be bred from fertile material like U-238 and Th-232 via neutron capture. Future fusion power plants, with total neutron source rates in exces...
We argue that the spatiotemporal dynamics of bred vectors in chaotic extended systems are related to a kinetic roughening process in the Kardar-Parisi-Zhang universality class. This implies that there exists a characteristic length scale correspond...
The breeding method is a computationally cheap way to generate flow-adapted ensembles to be used in probabilistic forecasts. Its main disadvantage is that the ensemble may lack diversity and collapse to a low-dimensional subspace. To still benefit fr...
Teaching digital manufacturing at scale using MOOCs has opened opportunities for IMT, a network of French graduate engineering schools, to work closely with a community of learners and educators in physical spaces called Fab Labs. By setting up a coh...
Magnetic Resonance Imaging is one of the most versatile experimental techniques in chemistry, physics and biology, providing insight into the structure and dynamics of matter at the molecular scale. A group led by Klaas Pruessmann at ETH Zurich (Br...
Continuum and emission-line variability of active galactic nuclei provides a powerful probe of microarcsecond scale structures in the central regions of these sources. In this contribution, we review basic concepts and methodologies used in analyzi...