otx.core.utils.cache

Cache Class for Trainer kwargs.

Classes

TrainerArgumentsCache(**kwargs)

Cache arguments.

class otx.core.utils.cache.TrainerArgumentsCache(**kwargs)

Bases: object

Cache arguments.

Since the Engine class accepts PyTorch Lightning Trainer arguments, we store these arguments using this class before the trainer is instantiated.

Parameters:

(**kwargs) –

Trainer arguments that are cached

Example

>>> conf = OmegaConf.load("config.yaml")
>>> cache =  TrainerArgumentsCache(**conf)
>>> cache.args
{
    ...
    'max_epochs': 100,
    'val_check_interval': 0
}
>>> config = {"max_epochs": 1, "val_check_interval": 1.0}
>>> cache.update(config)
Overriding max_epochs from 100 with 1
Overriding val_check_interval from 0 with 1.0
>>> cache.args
{
    ...
    'max_epochs': 1,
    'val_check_interval': 1.0
}

static get_trainer_constructor_args() → set[str]

Get the set of arguments accepted by the Trainer class constructor.

Returns:

A set of argument names accepted by the Trainer class constructor.

Return type:

set[str]

requires_update(**kwargs) → bool

Checks if the cached arguments need to be updated based on the provided keyword arguments.

Parameters:

**kwargs – The keyword arguments to compare with the cached arguments.

Returns:

True if any of the cached arguments need to be updated, False otherwise.

Return type:

bool

update(**kwargs) → None

Replace cached arguments with arguments retrieved from the model.

property args: dict[str, Any]

Returns the cached arguments.

Returns:

The cached arguments.

Return type:

dict[str, Any]