memorax.algorithms.PPOConfig

class memorax.algorithms.PPOConfig

Bases: object

PPOConfig(name: str, num_envs: int, num_eval_envs: int, num_steps: int, gamma: float, gae_lambda: float, num_minibatches: int, update_epochs: int, normalize_advantage: bool, clip_coef: float, clip_vloss: bool, ent_coef: float, vf_coef: float, target_kl: Optional[float] = None, burn_in_length: int = 0)

name: str

num_envs: int

num_eval_envs: int

num_steps: int

gamma: float

gae_lambda: float

num_minibatches: int

update_epochs: int

normalize_advantage: bool

clip_coef: float

clip_vloss: bool

ent_coef: float

vf_coef: float

target_kl: float | None = None

burn_in_length: int = 0

property batch_size

__init__(name, num_envs, num_eval_envs, num_steps, gamma, gae_lambda, num_minibatches, update_epochs, normalize_advantage, clip_coef, clip_vloss, ent_coef, vf_coef, target_kl=None, burn_in_length=0)

Parameters:

• name (str)

• num_envs (int)

• num_eval_envs (int)

• num_steps (int)

• gamma (float)

• gae_lambda (float)

• num_minibatches (int)

• update_epochs (int)

• normalize_advantage (bool)

• clip_coef (float)

• clip_vloss (bool)

• ent_coef (float)

• vf_coef (float)

• target_kl (float | None)

• burn_in_length (int)

Return type:

None

replace(**updates)

Returns a new object replacing the specified fields with new values.