tf_agents.replay_buffers.reverb_utils.ReverbAddEpisodeObserver

Observer for writing episodes to the Reverb replay buffer.

tf_agents.replay_buffers.reverb_utils.ReverbAddEpisodeObserver(
    py_client: tf_agents.typing.types.ReverbClient,
    table_name: Union[Text, Sequence[Text]],
    max_sequence_length: int,
    priority: Union[float, int] = 1,
    bypass_partial_episodes: bool = False
)

Args
py_client	Python client for the reverb replay server.
table_name	The table name(s) where samples will be written to.
max_sequence_length	An integer. max_sequence_length used to write to the replay buffer tables. This defines the size of the internal buffer controlling the upper limit of the number of timesteps which can be referenced in a single prioritized item. Note that this is the maximum number of trajectories across all the cached episodes that you are writing into the replay buffer (e.g. number_of_episodes). max_sequence_length is not a limit of how many timesteps or items that can be inserted into the replay buffer. Note that, since max_sequence_length controls the size of internal buffer, it is suggested not to set this value to a very large number. If the number of steps in an episode is more than max_sequence_length, only items up to max_sequence_length is written into the table.
priority	Initial priority for the table.
bypass_partial_episodes	If False (default) and an episode length is greater than max_sequence_length, a ValueError is raised. If set to True, the episodes with length more than max_sequence_length do not cause a ValueError. These episodes are bypassed (will NOT be written into the replay buffer) and an error message is shown to the user. Note that in this case (bypass_partial_episodes=True), the steps for episodes with length more than max_sequence_length are wasted and thrown away. This decision is made to guarantee that the replay buffer always has FULL episodes. Note that, max_sequence_length is just an upper bound.

Raises
ValueError	If priority is not numeric.
ValueError	If max_sequence_length is not positive.

Methods

close

View source

close()

Closes the writer of the observer.

Note: Using the observer after closing it is not supported.

open

View source

open()

Open the writer of the observer.

reset

View source

reset()

Resets the state of the observer.

Note that the data cached in the writer will NOT get automatically written into the Reverb table. If you wish to write the cached partial episode as a new sequences, call write_cached_steps instead.

update_priority

View source

update_priority(
    priority: Union[float, int]
)

Update the table priority.

Args
priority	Updates the priority of the observer.

ValueError: If priority is not numeric.

write_cached_steps

View source

write_cached_steps()

Writes the cached steps into the writer.

Note: The method resets the number of episodes and steps after writing the data into the replay buffer.

__call__

View source

__call__(
    trajectory
)

Writes the trajectory into the underlying replay buffer.

Allows trajectory to be a flattened trajectory. No batch dimension allowed.

Args
trajectory	The trajectory to be written which could be (possibly nested) trajectory object or a flattened version of a trajectory. It assumes there is no batch dimension.

Raises
ValueError	If bypass_partial_episodes == False and episode length is > max_sequence_length.