With distribution="normal"
, samples are drawn from a truncated normal
distribution centered on zero, with stddev = sqrt(scale / n)
where n is:
number of input units in the weight tensor, if mode = "fan_in"
number of output units, if mode = "fan_out"
average of the numbers of input and output units, if mode = "fan_avg"
initializer_variance_scaling(
scale = 1,
mode = c("fan_in", "fan_out", "fan_avg"),
distribution = c("normal", "uniform", "truncated_normal", "untruncated_normal"),
seed = NULL
)
Scaling factor (positive float).
One of "fan_in", "fan_out", "fan_avg".
One of "truncated_normal", "untruncated_normal" and "uniform". For backward compatibility, "normal" will be accepted and converted to "untruncated_normal".
Integer used to seed the random generator.
With distribution="uniform"
, samples are drawn from a uniform distribution
within -limit, limit
, with limit = sqrt(3 * scale / n)
.
Other initializers:
initializer_constant()
,
initializer_glorot_normal()
,
initializer_glorot_uniform()
,
initializer_he_normal()
,
initializer_he_uniform()
,
initializer_identity()
,
initializer_lecun_normal()
,
initializer_lecun_uniform()
,
initializer_ones()
,
initializer_orthogonal()
,
initializer_random_normal()
,
initializer_random_uniform()
,
initializer_truncated_normal()
,
initializer_zeros()