Tensor.glort_uniform()
A method for weight initialization of a tensor with a uniform distribution.
Paper link (opens in a new tab)
Usage
from tinygrad.tensor import Tensor
tensor = Tensor.glorot_uniform(3, 3)
print(tensor.numpy())
Return value
[[ 4.01388645e-01 -1.20162964e-04 -4.05807972e-01]
[-7.71492124e-01 -4.59152102e-01 1.59986973e-01]
[-1.15628719e-01 4.85111117e-01 8.84148717e-01]]