BaseSb3AlgorithmArgs
Full path:
schola.scripts.sb3.settings.BaseSb3AlgorithmArgs
schola.scripts.sb3.settings.BaseSb3AlgorithmArgs
BaseSb3AlgorithmArgs
BaseSb3AlgorithmArgs(learning_rate: 'Annotated[float, Parameter(validator=validators.Number(gt=0.0))]' = 0.0003, n_steps: 'Annotated[int, Parameter(validator=validators.Number(gte=1))]' = 2048, batch_size: 'Annotated[int, Parameter(validator=validators.Number(gte=1))]' = 64)Bases: object
Methods
| Item | Description |
|---|---|
| init | — |
Attributes
| Item | Description |
|---|---|
| batch_size | Minibatch size for each update. |
| learning_rate | Learning rate for the optimizer. |
| n_steps | Number of steps to run for each environment per update. |
Parameters
learning_rate (Annotated[float, Parameter(validator=(Number(lt=None, lte=None, gt=0, gte=None, modulo=None),))])
n_steps (Annotated[int, Parameter(validator=(Number(lt=None, lte=None, gt=None, gte=1, modulo=None),))])
batch_size (Annotated[int, Parameter(validator=(Number(lt=None, lte=None, gt=None, gte=1, modulo=None),))])
init
__init__(learning_rate=0.0003, n_steps=2048, batch_size=64)Parameters
learning_rate (Annotated[float, Parameter(validator=(Number(lt=None, lte=None, gt=0, gte=None, modulo=None),))])
n_steps (Annotated[int, Parameter(validator=(Number(lt=None, lte=None, gt=None, gte=1, modulo=None),))])
batch_size (Annotated[int, Parameter(validator=(Number(lt=None, lte=None, gt=None, gte=1, modulo=None),))])
Returns
None
batch_size
= 64 batch_size: Annotated[int, Parameter(validator=Number(lt=None, lte=None, gt=None, gte=1, modulo=None))]Minibatch size for each update. This is the number of timesteps used in each batch for training the policy. Must be a divisor of n_steps.
learning_rate
= 0.0003 learning_rate: Annotated[float, Parameter(validator=Number(lt=None, lte=None, gt=0, gte=None, modulo=None))]Learning rate for the optimizer.
n_steps
= 2048 n_steps: Annotated[int, Parameter(validator=Number(lt=None, lte=None, gt=None, gte=1, modulo=None))]Number of steps to run for each environment per update. This is the number of timesteps collected before updating the policy.