Scaling Vision Transformers to 22 Billion Parameters.
Mostafa DehghaniJosip DjolongaBasil MustafaPiotr PadlewskiJonathan HeekJustin GilmerAndreas SteinerMathilde CaronRobert GeirhosIbrahim AlabdulmohsinRodolphe JenattonLucas BeyerMichael TschannenAnurag ArnabXiao WangCarlos RiquelmeMatthias MindererJoan PuigcerverUtku EvciManoj KumarSjoerd van SteenkisteGamaleldin F. ElsayedAravindh MahendranFisher YuAvital OliverFantine HuotJasmijn BastingsMark Patrick CollierAlexey A. GritsenkoVighnesh BirodkarCristina VasconcelosYi TayThomas MensinkAlexander KolesnikovFilip PaveticDustin TranThomas KipfMario LucicXiaohua ZhaiDaniel KeysersJeremiah HarmsenNeil HoulsbyPublished in: CoRR (2023)
Keyphrases
- computer vision
- real time
- parameter selection
- parameter tuning
- vision system
- image processing
- neural network
- parameter settings
- real world
- expectation maximization
- parameter estimation
- em algorithm
- database
- visual perception
- operating conditions
- parameter values
- parameter space
- sensitivity analysis
- learning algorithm
- control system
- trade off
- object recognition
- website
- machine learning
- e learning