arlbench.core.algorithms.ppo.ppo

Proximal Policy Optimization (PPO) algorithm.

Classes

PPO(hpo_config, env[, eval_env, cnn_policy, ...])	JAX-based implementation of Proximal Policy Optimization (PPO).
PPOMetrics(loss, grads, advantages)	PPO metrics returned by train function.
PPORunnerState(rng, train_state, ...)	PPO runner state.
PPOState(runner_state[, buffer_state])	PPO algorithm state.
PPOTrainState(step, apply_fn, params, tx, ...)	PPO training state.
PPOTrainingResult(eval_rewards, ...)	PPO training result.
Transition(done, action, value, reward, ...)	PPO Transition.

class arlbench.core.algorithms.ppo.ppo.PPO(hpo_config, env, eval_env=None, cnn_policy=False, nas_config=None, deterministic_eval=True, track_trajectories=False, track_metrics=False)

Bases: Algorithm

JAX-based implementation of Proximal Policy Optimization (PPO).

static get_checkpoint_factory(runner_state, train_result)

Creates a factory dictionary of all posssible checkpointing options for PPO.

Parameters:

• runner_state (PPORunnerState) – Algorithm runner state.

• train_result (PPOTrainingResult | None) – Training result.

Returns:

Dictionary of factory functions containing [opt_state, params, loss, trajectories].

Return type:

dict[str, Callable]

static get_default_hpo_config()

Returns the default hyperparameter configuration for PPO.

Return type:

Configuration

static get_default_nas_config()

Returns the default neural architecture search configuration for PPO.

Return type:

Configuration

static get_hpo_config_space(seed=None)

Returns the hyperparameter configuration space for PPO.

Return type:

ConfigurationSpace

static get_nas_config_space(seed=None)

Returns the neural architecture search configuration space for PPO.

Return type:

ConfigurationSpace

init(rng, network_params=None, opt_state=None)

Initializes PPO state. Passed parameters are not initialized and included in the final state.

Parameters:

• rng (chex.PRNGKey) – Random generator key.

• network_params (FrozenDict | dict | None, optional) – Network parameters. Defaults to None.

• opt_state (optax.OptState | None, optional) – Optimizer state. Defaults to None.

Returns:

PPO state.

Return type:

PPOState

predict(runner_state, obs, rng, deterministic=True)

Predict action(s) based on the current observation(s).

Parameters:

• runner_state (PPORunnerState) – Algorithm runner state.

• obs (jnp.ndarray) – Observation(s).

• rng (chex.PRNGKey | None, optional) – Random generator key. Defaults to None.

• deterministic (bool) – Return deterministic action. Defaults to True.

Returns:

Action(s).

Return type:

jnp.ndarray

train(runner_state, _, n_total_timesteps=1000000, n_eval_steps=10, n_eval_episodes=10)

Performs one iteration of training.

Parameters:

• runner_state (PPORunnerState) – PPO runner state.

• _ (None) – Unused parameter (buffer_state in other algorithms).

• n_total_timesteps (int, optional) – Total number of training timesteps. Update steps = n_total_timesteps // n_envs. Defaults to 1000000.

• n_eval_steps (int, optional) – Number of evaluation steps during training. Defaults to 100.

• n_eval_episodes (int, optional) – Number of evaluation episodes per evaluation during training. Defaults to 10.

Returns:

Tuple of PPO algorithm state and training result.

Return type:

PPOTrainReturnT

class arlbench.core.algorithms.ppo.ppo.PPOMetrics(loss: jnp.ndarray, grads: jnp.ndarray | dict, advantages: jnp.ndarray)

Bases: NamedTuple

PPO metrics returned by train function. Consists of (loss, grads, advantages).

advantages: Array

Alias for field number 2

grads: Array | dict

Alias for field number 1

loss: Array

Alias for field number 0

class arlbench.core.algorithms.ppo.ppo.PPORunnerState(rng: chex.PRNGKey, train_state: PPOTrainState, normalizer_state: RunningStatisticsState, env_state: Any, obs: chex.Array, global_step: int)

Bases: NamedTuple

PPO runner state. Consists of (rng, train_state, env_state, obs, global_step).

env_state: Any

Alias for field number 3

global_step: int

Alias for field number 5

normalizer_state: RunningStatisticsState

Alias for field number 2

obs: Union[Array, ndarray, bool_, number]

Alias for field number 4

rng: Array

Alias for field number 0

train_state: PPOTrainState

Alias for field number 1

class arlbench.core.algorithms.ppo.ppo.PPOState(runner_state: PPORunnerState, buffer_state: None = None)

Bases: NamedTuple

PPO algorithm state. Consists of (runner_state, buffer_state).

Note: As PPO does not use a buffer buffer_state is always None and only kept for consistency across algorithms.

buffer_state: None

Alias for field number 1

runner_state: PPORunnerState

Alias for field number 0

class arlbench.core.algorithms.ppo.ppo.PPOTrainState(step, apply_fn, params, tx, opt_state)

Bases: TrainState

PPO training state.

classmethod create_with_opt_state(*, apply_fn, params, tx, opt_state, **kwargs)

Creates a PPO training state with an optimizer state.

replace(**updates)

“Returns a new object replacing the specified fields with new values.

class arlbench.core.algorithms.ppo.ppo.PPOTrainingResult(eval_rewards: jnp.ndarray, trajectories: Transition | None, metrics: PPOMetrics | None)

Bases: NamedTuple

PPO training result. Consists of (eval_rewards, trajectories, metrics).

eval_rewards: Array

Alias for field number 0

metrics: PPOMetrics | None

Alias for field number 2

trajectories: Transition | None

Alias for field number 1

class arlbench.core.algorithms.ppo.ppo.Transition(done: jnp.ndarray, action: jnp.ndarray, value: jnp.ndarray, reward: jnp.ndarray, log_prob: jnp.ndarray, obs: jnp.ndarray, info: dict)

Bases: NamedTuple

PPO Transition. Consists of (done, action, value, reward, log_prob, obs, info).

action: Array

Alias for field number 1

done: Array

Alias for field number 0

info: dict

Alias for field number 6

log_prob: Array

Alias for field number 4

obs: Array

Alias for field number 5

reward: Array

Alias for field number 3

value: Array

Alias for field number 2

previous

arlbench.core.algorithms.ppo.models

next

arlbench.core.algorithms.prioritised_item_buffer