SCHENLIU / longformer-chinese

chinese version of longformer
110 stars 15 forks source link

预训练参数和耗时 #7

Open xuehui0725 opened 3 years ago

xuehui0725 commented 3 years ago

请问下,你预训练长度4096,是单卡GPU还是多卡GPU进行训练的,如果是单卡,显存是多大的呢?预训练的batch_size设置了多大,整个预训练过程用了多久呢?