Results 61 to 70 of about 27,399 (256)
WAPS-Quant: Low-Bit Post-Training Quantization Using Weight-Activation Product Scaling
Post-Training Quantization (PTQ) has been effectively compressing neural networks into very few bits using a limited calibration dataset. Various quantization methods utilizing second-order error have been proposed and demonstrated good performance ...
Geunjae Choi, Kamin Lee, Nojun Kwak
doaj +1 more source
Periodicity and Transport from Round-Off Errors [PDF]
We investigate the effects of round-off errors on quasi-periodic motions in a linear symplectic planar map. By discretizing coordinates uniformly we transform this map into a permutation of Z2, and study motions near infinity, which correspond to a fine discretization. We provide numerical evidence that all orbits are periodic and that the averageorder
openaire +2 more sources
Postbuild annealing systematically modifies the phase fractions and morphology of EB‐PBF processed Mo–9Si–8B. Quantitative microstructure–property correlations reveal how controlled phase evolution enhances high‐temperature compressive strength and creep resistance.
Christopher Schmidt +5 more
wiley +1 more source
Feasibility and Limitations of the Advanced Replica Process for Bioactive Glasses
Periodic 3D‐printed Kelvin‐cell templates enable a controlled advanced replica process for fabricating highly porous glass lattices. Across P45K9, 45S5, and 13–93, material‐dependent differences in coating, burnout, and shrinkage lead to distinct strut morphologies, while comparable final porosities of 83%–88% are retained.
Swantje Funk +4 more
wiley +1 more source
Camparison of Numerically Stability of Two Algorithms for the Calculation of Variance [PDF]
In descriptive statistics, there are two computational algorithms for determining the variance S2, of a set of observations : Algorithm 1: S2= - , Algorithm 2: S2= , where .
D Rostami
doaj
This perspective reframes additive manufacturing for electrical machines as a qualification‐limited materials and architecture design problem. It links process–structure–property–performance relationships to magnetic, conducting, dielectric, and thermal property windows, highlighting where AM can enable segmented magnetic circuits, permanent magnet ...
Dénes Fodor, Loránd Szabó
wiley +1 more source
Support and Optimization of Multi-Granularity Quantization Framework for Deep Learning Compiler [PDF]
With the increasing demand for the deployment of large models by major manufacturers, the accuracy of the single quantization method of deep learning compiler Tensor Virtual Machine (TVM) has decreased, and this method is no longer sufficient to satisfy ...
WEI Mingkang, LI Jianan, HAN Lin, GAO Wei, ZHAO Rongcai, WANG Hongsheng
doaj +1 more source
Additive manufacturing introduces pronounced microstructural and crystallographic anisotropy in Ni‐based superalloys, resulting in orientation‐dependent creep deformation and damage. This review identifies the key microstructural origins of creep anisotropy, examines their influence on dominant deformation mechanisms, and highlights processing ...
Chandan Kumar +3 more
wiley +1 more source
Impact of Hydrogen Incorporation on Electronic and Magnetic Structure of X2CrNi18‐9 Stainless Steel
Hydrogen uptake in austenitic stainless steel alters both its electronic and magnetic structure. Seebeck measurements reveal hydrogen‐induced shifts in the electronic density of states, while neutron scattering uncovers nanoscale inhomogeneities that grow upon hydrogen incorporation, with the most pronounced effect observed in the PBF as‐built state ...
Torben Tappe +9 more
wiley +1 more source
Laser‐Induced Graphene from Waste Almond Shells
Almond shells, an abundant agricultural by‐product, are repurposed to create a fully bioderived almond shell/chitosan composite (ASC) degradable in soil. ASC is converted into laser‐induced graphene (LIG) by laser scribing and proposed as a substrate for transient electronics.
Yulia Steksova +9 more
wiley +1 more source

