`evorl.algorithms.erl.cemrl_td3.cemrl_origin`¶

Module Contents¶

Classes¶

`CEMRLTrainMetric`
`CEMRLWorkflow`	Original CEMRL impl.
`WorkflowMetric`

API¶

class evorl.algorithms.erl.cemrl_td3.cemrl_origin.CEMRLTrainMetric[source]¶

Bases: evorl.algorithms.erl.cemrl_workflow.CEMRLTrainMetric

num_updates_per_iter: chex.Array¶: ‘zeros(…)’

time_cost_per_iter: float¶: 0.0

class evorl.algorithms.erl.cemrl_td3.cemrl_origin.CEMRLWorkflow(**kwargs)[source]¶

Bases: evorl.algorithms.erl.cemrl_td3.cemrl.CEMRLWorkflow

Original CEMRL impl.

1 critic + n actors + 1 replay buffer.

classmethod enable_jit() → None[source]¶

classmethod name()[source]¶

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

class evorl.algorithms.erl.cemrl_td3.cemrl_origin.WorkflowMetric[source]¶

Bases: evorl.metrics.MetricBase

iterations: chex.Array¶: ‘zeros(…)’

sampled_episodes: chex.Array¶: ‘zeros(…)’

sampled_timesteps: chex.Array¶: ‘zeros(…)’

sampled_timesteps_per_iter: chex.Array¶: ‘zeros(…)’