megengine.random.distribution.
normal
服从高斯分布的随机变量 \(N(\mu, \sigma)\)
size (Optional[Iterable[int]]) – 输出张量的形状
Optional
Iterable
int
mean (float) – 分布的期望(平均数)
float
std (float) – 分布的标准差 (variance = \(\sigma ^ 2\))
Tensor
输出张量
示例代码:
import megengine as mge import megengine.random as rand x = rand.normal(mean=0, std=1, size=(2, 2)) print(x.numpy())
输出:
[[-0.20235455 -0.6959438 ] [-1.4939808 -1.5824696 ]]
uniform
服从均匀分布 $U(0, 1)$ 的随机变量
low (float) – 范围下界
high (float) – 范围上界
import megengine as mge import megengine.random as rand x = rand.uniform(size=(2, 2)) print(x.numpy())
[[0.76901674 0.70496535] [0.09365904 0.62957656]]
megengine.random.rng.
seed