evorl.algorithms.ddpg

Module Contents

Classes

DDPGAgent	The Agnet for DDPG.
DDPGNetworkParams	Contains training state for the learner.
DDPGTrainMetric
DDPGWorkflow

Functions

make_mlp_ddpg_agent

API

class evorl.algorithms.ddpg.DDPGAgent

Bases: evorl.agent.Agent

The Agnet for DDPG.

actor_loss(agent_state: evorl.agent.AgentState, sample_batch: evorl.sample_batch.SampleBatch, key: chex.PRNGKey) → evorl.types.LossDict

actor_network: flax.linen.Module

None

compute_actions(agent_state: evorl.agent.AgentState, sample_batch: evorl.sample_batch.SampleBatch, key: chex.PRNGKey) → tuple[evorl.types.Action, evorl.types.PolicyExtraInfo]

critic_loss(agent_state: evorl.agent.AgentState, sample_batch: evorl.sample_batch.SampleBatch, key: chex.PRNGKey) → evorl.types.LossDict

critic_network: flax.linen.Module

None

discount: float

1

evaluate_actions(agent_state: evorl.agent.AgentState, sample_batch: evorl.sample_batch.SampleBatch, key: chex.PRNGKey) → tuple[evorl.types.Action, evorl.types.PolicyExtraInfo]

exploration_epsilon: float

0.5

init(obs_space: evorl.envs.Space, action_space: evorl.envs.Space, key: chex.PRNGKey) → evorl.agent.AgentState

property normalize_obs

obs_preprocessor: Any

‘pytree_field(…)’

class evorl.algorithms.ddpg.DDPGNetworkParams

Bases: evorl.types.PyTreeData

Contains training state for the learner.

actor_params: evorl.types.Params

None

critic_params: evorl.types.Params

None

target_actor_params: evorl.types.Params

None

target_critic_params: evorl.types.Params

None

class evorl.algorithms.ddpg.DDPGTrainMetric

Bases: evorl.metrics.MetricBase

actor_loss: chex.Array

None

critic_loss: chex.Array

None

raw_loss_dict: evorl.types.LossDict

‘metric_field(…)’

class evorl.algorithms.ddpg.DDPGWorkflow(env: evorl.envs.Env, agent: evorl.agent.Agent, optimizer: optax.GradientTransformation, evaluator: evorl.evaluators.Evaluator, replay_buffer: evorl.replay_buffers.AbstractReplayBuffer, config: omegaconf.DictConfig)

Bases: evorl.algorithms.offpolicy_utils.OffPolicyWorkflowTemplate

classmethod name()

step(state: evorl.types.State) → tuple[evorl.metrics.MetricBase, evorl.types.State]

evorl.algorithms.ddpg.make_mlp_ddpg_agent(action_space: evorl.envs.Space, critic_hidden_layer_sizes: tuple[int] = (256, 256), actor_hidden_layer_sizes: tuple[int] = (256, 256), discount: float = 1, exploration_epsilon: float = 0.5, normalize_obs: bool = False, policy_obs_key: str = '', value_obs_key: str = '')