one epoch: 指所有training samples的一次forward pass和一次backward pass.
batch size: 指一次forward/backward pass中training samples的数量.batch size越大,所需的内存空间越多.
iterations: number of passes, each pass using [batch size] number of examples.
one pass = one forward pass + one backward pass
例:对于1000个训练样本,batch size是500,那么每个epoch需要迭代两次.