Dropout layer

Dropout layer

Usage

mlx_dropout(p = 0.5)

Arguments

p

Probability of dropping an element (default: 0.5).

Value

An mlx_module applying dropout during training.

See also

mlx.nn.Dropout

Examples

set.seed(1)
dropout <- mlx_dropout(p = 0.3)
x <- mlx_matrix(1:12, 3, 4)
mlx_forward(dropout, x)
#> mlx array [3 x 4]
#>   dtype: float32
#>   device: cpu
#>   values:
#>          [,1]     [,2]     [,3]     [,4]
#> [1,] 0.000000 0.000000 10.00000 14.28571
#> [2,] 2.857143 7.142857 11.42857 15.71429
#> [3,] 0.000000 8.571428  0.00000  0.00000