ML Community Day is November 9! Join us for updates from TensorFlow, JAX, and more Learn more

Module: tf_agents.trajectories

Trajectories module.

Modules

policy_step module: Policy Step.

time_step module: TimeStep representing a step in the environment.

trajectory module: Trajectory containing time_step transition information.

Classes

class PolicyInfo: PolicyInfo(log_probability,)

class PolicyStep: Returned with every call to policy.action() and policy.distribution().

class StepType: Defines the status of a TimeStep within a sequence.

class TimeStep: Returned with every call to step and reset on an environment.

class Trajectory: A tuple that represents a trajectory.

class Transition: A tuple that represents a transition.

Functions

boundary(...): Create a Trajectory transitioning between StepTypes LAST and FIRST.

first(...): Create a Trajectory transitioning between StepTypes FIRST and MID.

from_transition(...): Returns a Trajectory given transitions.

last(...): Create a Trajectory transitioning between StepTypes MID and LAST.

mid(...): Create a Trajectory transitioning between StepTypes MID and MID.

restart(...): Returns a TimeStep with step_type set equal to StepType.FIRST.

single_step(...): Create a Trajectory transitioning between StepTypes FIRST and LAST.

termination(...): Returns a TimeStep with step_type set to StepType.LAST.

time_step_spec(...): Returns a TimeStep spec given the observation_spec.

to_n_step_transition(...): Create an n-step transition from a trajectory with T=N + 1 frames.

to_transition(...): Create a transition from a trajectory or two adjacent trajectories.

to_transition_spec(...): Create a transition spec from a trajectory spec.

transition(...): Returns a TimeStep with step_type set equal to StepType.MID.

truncation(...): Returns a TimeStep with step_type set to StepType.LAST.