evorl.algorithms.td7

Module Contents

Classes

TD7Actor
TD7Agent	The Agent for TD7.
TD7Critic
TD7Encoder
TD7NetworkParams
TD7TrainMetric
TD7Workflow

Functions

avg_l1_norm	Average L1 Norm used in TD7.
make_td7_agent

API

class evorl.algorithms.td7.TD7Actor

Bases: flax.linen.Module

action_size: int

None

hidden_layer_sizes: Sequence[int]

(256, 256)

state_emb_dim: int

256

z_s_dim: int

256

class evorl.algorithms.td7.TD7Agent

Bases: evorl.agent.Agent

The Agent for TD7.

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

clip_policy_noise: float

0.5

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

0.99

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

encoder_network: flax.linen.Module

None

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

0.1

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

min_priority: float

1.0

property normalize_obs

obs_preprocessor: Any

‘pytree_field(…)’

policy_noise: float

0.2

class evorl.algorithms.td7.TD7Critic

Bases: flax.linen.Module

hidden_layer_sizes: Sequence[int]

(256, 256)

state_action_emb_dim: int

256

z_s_dim: int

256

z_sa_dim: int

256

class evorl.algorithms.td7.TD7Encoder

Bases: flax.linen.Module

f_layer_sizes: Sequence[int]

(256, 256)

g_layer_sizes: Sequence[int]

(256, 256)

setup()

z_s_dim: int

256

z_sa_dim: int

256

zs(obs: jax.Array) → jax.Array

zsa(z_s: jax.Array, action: jax.Array) → jax.Array

class evorl.algorithms.td7.TD7NetworkParams

Bases: evorl.types.PyTreeData

actor_params: evorl.types.Params

None

checkpoint_actor_params: evorl.types.Params

None

checkpoint_encoder_params: evorl.types.Params

None

critic_params: evorl.types.Params

None

encoder_params: evorl.types.Params

None

fixed_encoder_params: evorl.types.Params

None

fixed_encoder_target_params: evorl.types.Params

None

target_actor_params: evorl.types.Params

None

target_critic_params: evorl.types.Params

None

class evorl.algorithms.td7.TD7TrainMetric

Bases: evorl.metrics.MetricBase

actor_loss: chex.Array

None

critic_loss: chex.Array

None

encoder_loss: chex.Array

None

raw_loss_dict: evorl.types.LossDict

‘metric_field(…)’

class evorl.algorithms.td7.TD7Workflow(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

learn(state: evorl.types.State) → evorl.types.State

classmethod name()

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

evorl.algorithms.td7.avg_l1_norm(x: jax.Array, eps: float = 1e-08) → jax.Array

Average L1 Norm used in TD7.

evorl.algorithms.td7.make_td7_agent(action_space: evorl.envs.Space, z_s_dim: int = 256, z_sa_dim: int = 256, f_layer_sizes: Sequence[int] = (256, 256), g_layer_sizes: Sequence[int] = (256, 256), state_emb_dim: int = 256, state_action_emb_dim: int = 256, critic_hidden_layer_sizes: Sequence[int] = (256, 256), actor_hidden_layer_sizes: Sequence[int] = (256, 256), discount: float = 0.99, exploration_epsilon: float = 0.1, policy_noise: float = 0.2, clip_policy_noise: float = 0.5, min_priority: float = 1.0, normalize_obs: bool = False)