6
@zhuzaileiting 你这是用cpu训练的,GPU速度大概在1.3-2 steps/s
Originally posted by @miven in https://github.com/babysor/MockingBird/issues/20#issuecomment-903427845
@zhuzaileiting 你这是用cpu训练的,GPU速度大概在1.3-2 steps/s
Originally posted by @miven in https://github.com/babysor/MockingBird/issues/20#issuecomment-903427845
为什么提高batch_size时,速度反而降了
3080ti 用 cuda12.2 每秒 1.1steps
3080ti 用 cuda12.2 每秒 1.1steps
速度很慢,而且将batch_size提高会降低速度,cpu会卡顿