diracnet.py 文件源码

def batch_norm(x, params, stats, base, mode):
    return F.batch_norm(x, weight=params[base + '.weight'],
                        bias=params[base + '.bias'],
                        running_mean=stats[base + '.running_mean'],
                        running_var=stats[base + '.running_var'],
                        training=mode)