evorl.algorithms.meta.pbt_utils

Module Contents

Functions

convert_pop_to_df
exp_uniform_init	Sample from exp(-x).
log_uniform_init	Random sample from log space.
truncated_normal_init	Random sample from a truncated normal distribution.
uniform_init	Random sample.

API

evorl.algorithms.meta.pbt_utils.convert_pop_to_df(pop)

evorl.algorithms.meta.pbt_utils.exp_uniform_init(search_space, key: chex.PRNGKey, num: int) → chex.Array

Sample from exp(-x).

evorl.algorithms.meta.pbt_utils.log_uniform_init(search_space, key: chex.PRNGKey, num: int) → chex.Array

Random sample from log space.

Suitable for hyperparameters that need explore different magnitudes in positive range. eg: [1e-3, 100].

evorl.algorithms.meta.pbt_utils.truncated_normal_init(search_space, key: chex.PRNGKey, num: int, m=0.95) → chex.Array

Random sample from a truncated normal distribution.

evorl.algorithms.meta.pbt_utils.uniform_init(search_space, key: chex.PRNGKey, num: int) → chex.Array

Random sample.