PyTorch - Difference between rand() and randn()

rand() - Returns a tensor filled with random numbers from a uniform distribution on the interval [0, 1)

torch.rand(100)

 

tensor([0.7880, 0.3032, 0.3627, 0.7082, 0.1795, 0.4985, 0.9594, 0.7900, 0.9585,
        0.9081, 0.1518, 0.7774, 0.5773, 0.7038, 0.6504, 0.5719, 0.1926, 0.9394,
        0.6213, 0.8958, 0.4246, 0.5511, 0.9647, 0.1502, 0.7306, 0.8174, 0.7219,
        0.3699, 0.6787, 0.8045, 0.1807, 0.6873, 0.7628, 0.2076, 0.3093, 0.9894,
        0.0346, 0.4468, 0.5628, 0.4541, 0.2028, 0.0766, 0.9851, 0.0201, 0.7290,
        0.2190, 0.5702, 0.4273, 0.9675, 0.5790, 0.0305, 0.1053, 0.2790, 0.2650,
        0.1109, 0.9617, 0.1193, 0.1121, 0.8944, 0.0246, 0.7383, 0.6698, 0.6169,
        0.9427, 0.0242, 0.1348, 0.8053, 0.8895, 0.4818, 0.1180, 0.3640, 0.5278,
        0.6612, 0.6710, 0.4686, 0.3391, 0.3898, 0.8033, 0.8198, 0.2833, 0.8570,
        0.1580, 0.6899, 0.1758, 0.3869, 0.7366, 0.9442, 0.1048, 0.6854, 0.1650,
        0.9399, 0.5242, 0.9021, 0.1574, 0.4322, 0.0012, 0.4553, 0.1467, 0.4122,
        0.6095])

 

 

randn() - Returns a tensor filled with random numbers from a normal distribution with mean 0 and variance 1 (also called the standard normal distribution).

dummy_randn = torch.randn(100)
dummy_randn

 

tensor([ 0.2083, -0.1248,  1.6377,  0.4158, -0.7428, -0.4556,  0.0847,  0.6920,
        -0.5236,  2.8119,  0.7568,  1.6511,  0.9284, -0.7238,  0.8877, -1.5698,
         0.6462, -0.4516,  0.4115, -0.5270, -0.9781,  0.1980,  1.8830, -0.1075,
        -2.4378,  0.3186, -0.3314, -0.5624,  0.9069, -1.4985, -0.4653, -0.8702,
        -1.5003,  0.9597,  1.0643,  1.1367,  0.2547,  1.4202, -0.7661, -0.2886,
        -0.4551, -0.7584, -1.8408,  0.1006, -0.8138,  0.9638, -1.1081,  1.2978,
        -1.1531, -1.4703, -0.2286, -0.0533,  0.1332,  1.4026,  1.3106,  0.2118,
         0.3589,  0.8719, -0.5595, -1.2247, -1.0900,  0.1247, -0.2331, -0.1365,
        -0.4995,  0.5812,  0.4895,  1.4686, -0.4431,  1.4738, -0.7067,  1.2377,
         0.8108, -1.5416, -0.0712, -0.1495, -0.0450, -1.3758,  1.6389, -0.3777,
        -0.5939,  0.5944,  0.8223, -1.8278,  1.3142, -0.7945,  0.7133, -0.4062,
         0.0681,  1.4935, -0.8036, -0.6210,  1.1735, -2.2997, -0.2382, -1.9110,
         0.1857,  1.2263,  0.2231, -1.6662])

 

dummy_randn.mean(), dummy_randn.var()

# (tensor(-0.0086), tensor(1.0671))