SNN2.src.model.replayMemory.mem module

ReplayMemory module

Class used to manage a replay memory

This module cannot be used untill an incomability between TF Agents and keras is solved in order to use tf_uniform_replay_buffer from TF Agents.

class SNN2.src.model.replayMemory.mem.ReplayMemory(data_spec: Tuple[TensorSpec, ...], max_length: int, utilization: int | None = None, output_file: str | None = None, logger: LogHandler | None = None, batch_size: int = 1)

Bases: object

append(trajectories: Tuple[Tensor, ...])

clear() → None

dump(*args, **kwargs)

take(n: int | None = None, batch_size: int | None = 1)

write_msg(msg: str, level: int = 20) → None

__write_msg. write a message into the log file with a defined log level

Parameters:

• msg (str) – msg to print

• level (int) – level default info

Return type:

none