trinity.algorithm.sample_strategy package

Submodules

• trinity.algorithm.sample_strategy.mix_sample_strategy module
  • MixSampleStrategy
    • MixSampleStrategy.__init__()
    • MixSampleStrategy.sample()
    • MixSampleStrategy.default_args()
    • MixSampleStrategy.state_dict()
    • MixSampleStrategy.load_state_dict()
• trinity.algorithm.sample_strategy.sample_strategy module
  • SampleStrategy
    • SampleStrategy.__init__()
    • SampleStrategy.set_model_version_metric()
    • SampleStrategy.sample()
    • SampleStrategy.default_args()
    • SampleStrategy.state_dict()
    • SampleStrategy.load_state_dict()
  • DefaultSampleStrategy
    • DefaultSampleStrategy.__init__()
    • DefaultSampleStrategy.sample()
    • DefaultSampleStrategy.default_args()
    • DefaultSampleStrategy.state_dict()
    • DefaultSampleStrategy.load_state_dict()
  • StalenessControlSampleStrategy
    • StalenessControlSampleStrategy.__init__()
    • StalenessControlSampleStrategy.sample()
  • WarmupSampleStrategy
    • WarmupSampleStrategy.__init__()
• trinity.algorithm.sample_strategy.utils module
  • representative_sample()

Module contents

class trinity.algorithm.sample_strategy.SampleStrategy(buffer_config: BufferConfig, **kwargs)

Bases: ABC

__init__(buffer_config: BufferConfig, **kwargs) → None

set_model_version_metric(exp_list: List[Experience], metrics: Dict)

abstract async sample(step: int) → Tuple[Experiences, Dict, List]

Sample data from buffer.

Parameters:

step (int) – The step number of current step.

Returns:

The sampled Experiences data. Dict: Metrics for logging. List: Representative data for logging.

Return type:

Experiences

abstract classmethod default_args() → dict

Get the default arguments of the sample strategy.

abstract state_dict() → dict

Get the state dict of the sample strategy.

abstract load_state_dict(state_dict: dict) → None

Load the state dict of the sample strategy.