d3rlpy.optimizers.AdamFactory

class d3rlpy.optimizers.AdamFactory(clip_grad_norm=None, lr_scheduler_factory=None, betas=(0.9, 0.999), eps=1e-08, weight_decay=0, amsgrad=False)

An alias for Adam optimizer.

from d3rlpy.optimizers import AdamFactory

factory = AdamFactory(weight_decay=1e-4)

Parameters:

• clip_grad_norm (Optional[float]) – Maximum norm value of gradients to clip.

• lr_scheduler_factory (Optional[LRSchedulerFactory]) – LRSchedulerFactory.

• betas (tuple[float, float]) – coefficients used for computing running averages of gradient and its square.

• eps (float) – term added to the denominator to improve numerical stability.

• weight_decay (float) – weight decay (L2 penalty).

• amsgrad (bool) – flag to use the AMSGrad variant of this algorithm.

Methods

create(named_modules, lr, compiled)

Returns an optimizer object.

Parameters:

• named_modules (list) – List of tuples of module names and modules.

• lr (float) – Learning rate.

• compiled (bool) – Flag to be True if CudaGraph and torch.compile are applied.

Returns:

OptimizerWrapper object.

Return type:

OptimizerWrapper

create_optimizer(named_modules, lr)

Parameters:

• named_modules (Iterable[tuple[str, torch.nn.modules.module.Module]]) –

• lr (float) –

Return type:

Adam

classmethod deserialize(serialized_config)

Parameters:

serialized_config (str) –

Return type:

TConfig

classmethod deserialize_from_dict(dict_config)

Parameters:

dict_config (dict[str, Any]) –

Return type:

TConfig

classmethod deserialize_from_file(path)

Parameters:

path (str) –

Return type:

TConfig

classmethod from_dict(kvs, *, infer_missing=False)

Parameters:

kvs (Optional[Union[dict, list, str, int, float, bool]]) –

Return type:

A

classmethod from_json(s, *, parse_float=None, parse_int=None, parse_constant=None, infer_missing=False, **kw)

Parameters:

s (Union[str, bytes, bytearray]) –

Return type:

A

static get_type()

Return type:

str

classmethod schema(*, infer_missing=False, only=None, exclude=(), many=False, context=None, load_only=(), dump_only=(), partial=False, unknown=None)

Parameters:

• infer_missing (bool) –

• many (bool) –

• partial (bool) –

Return type:

SchemaF[A]

serialize()

Return type:

str

serialize_to_dict()

Return type:

dict[str, Any]

to_dict(encode_json=False)

Return type:

Dict[str, Optional[Union[dict, list, str, int, float, bool]]]

to_json(*, skipkeys=False, ensure_ascii=True, check_circular=True, allow_nan=True, indent=None, separators=None, default=None, sort_keys=False, **kw)

Parameters:

• skipkeys (bool) –

• ensure_ascii (bool) –

• check_circular (bool) –

• allow_nan (bool) –

• indent (Optional[Union[int, str]]) –

• separators (Optional[Tuple[str, str]]) –

• default (Optional[Callable]) –

• sort_keys (bool) –

Return type:

str

Attributes

amsgrad: bool = False

betas: tuple[float, float] = (0.9, 0.999)

clip_grad_norm: Optional[float] = None

eps: float = 1e-08

lr_scheduler_factory: Optional[LRSchedulerFactory] = None

weight_decay: float = 0