4 papers
Mixed precision accumulation for neural network inference guided by componentwise forward error analysis
El-Mehdi El Arar, Silviu-Ioan Filip, Theo Mary +1
This work proposes a mathematically founded mixed precision accumulation strategy for the inference of neural networks. Our strategy is based on a new componentwise forward error a…
SwiftSketch: A Diffusion Model for Image-to-Vector Sketch Generation
Ellie Arar, Yarden Frenkel, Daniel Cohen-Or +2
Recent advancements in large vision-language models have enabled highly expressive and diverse vector sketch generation. However, state-of-the-art methods rely on a time-consuming…
Error Analysis of Sum-Product Algorithms under Stochastic Rounding
Pablo de Oliveira Castro, El-Mehdi El Arar, Eric Petit +1
The quality of numerical computations can be measured through their forward error, for which finding good error bounds is challenging in general. For several algorithms and using s…
The Positive Effects of Stochastic Rounding in Numerical Algorithms
El-Mehdi El Arar, Devan Sohier, Pablo de Oliveira Castro +1
Recently, stochastic rounding (SR) has been implemented in specialized hardware but most current computing nodes do not yet support this rounding mode. Several works empirically il…