首页 > TAG信息列表 > T2T

【报告笔记】基因组组装的最后挑战-T2T

目录长读长组装发展人类基因组T2T联盟CHM13-T2T基因组组装的最后挑战 长读长组装发展 2012:三代组装、二代校正;耗资源,适合小基因组,如细菌,4-15%错误率 2013:三代组装、三代校正;仍然只适用小的 2014:华夏一号(中国人三代参考基因组) 2016:Falcon/Falcon-Unzip,三代Pacbio二倍体真核生物 20

[2101] Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet

paper: arXiv code: pytorch main limitations of ViT straightforward tokenization of input images by hard split makes ViT unable to model local information, thus requiring more training samples than CNNs to achieve similar performanceself-attention in ViT