reference/generated/torchrl.modules.tensordict_module.zoo.MicroDuckSkills

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

MicroDuckSkills

class torchrl.modules.tensordict_module.zoo.MicroDuckSkills(policy: TensorDictModuleBase, task_library: MicroDuckTask, action_scale: float)

A deployable MicroDuck skill policy and its environment metadata.

policy maps a task-conditioned MicroDuck observation to normalized joint targets. task_library preserves the exact meaning and order of its task embeddings. action_scale records how the joint targets were applied during training. Keeping the three together prevents a high-level environment from silently pairing a policy with incompatible task ids or motor scaling.

Parameters:

• policy – Trained task-conditioned TensorDict policy.

• task_library – Ordered, stacked MicroDuckTask.

• action_scale – Environment-side joint-target scale used during training.

Examples

Download the pinned published skills and pass the resulting object to the high-level environment rather than unpacking policy metadata:

>>> from torchrl.modules.tensordict_module.zoo import MicroDuckSkills
>>> skills = MicroDuckSkills.from_pretrained()  
>>> skill_policy = skills.policy  
>>> task_library = skills.task_library  

Promote compatible joint-level dynamics to a high-level environment without separating the policy from that metadata:

>>> from torchrl.envs import MicroDuckSkillEnv
>>> base_env = make_microduck_game_env(  
...     action_scale=skills.action_scale
... )
>>> env = MicroDuckSkillEnv.from_env(  
...     base_env, skills, control_steps_per_decision=5
... )

See also

MicroDuckSkillPolicy is the neural policy stored here; MicroDuckTask describes one row of the ordered task library; MicroDuckEnv supplies the joint-level training dynamics; and MicroDuckSkillEnv deploys the complete artifact.

classmethod from_checkpoint(checkpoint: str | Path | Mapping[str, Any], *, device: device | str = 'cpu', freeze: bool = True, sha256: str | None = None) → MicroDuckSkills

Rebuild the policy and deployment metadata from a checkpoint.

Parameters:

• checkpoint – Local checkpoint path or an already loaded payload.

• device – Device for the rebuilt policy.

• freeze – Load in evaluation mode and disable gradients.

• sha256 – Expected SHA-256 digest for a path. This cannot be used with an already loaded payload.

classmethod from_pretrained(repo_id: str | None = None, *, filename: str | None = None, revision: str | None = None, device: device | str = 'cpu', freeze: bool = True, sha256: str | None = None, **hub_kwargs: Any) → MicroDuckSkills

Download the pinned published skills and rebuild them.

Parameters:

• repo_id – Hugging Face repository. Defaults to "torchrl/microduck-skills".

• filename – Checkpoint path in the repository. Defaults to the historical "walker.ckpt" artifact name.

• revision – Immutable repository revision. Defaults to the published six-skill policy revision.

• device – Device for the rebuilt policy.

• freeze – Load in evaluation mode and disable gradients.

• sha256 – Expected checkpoint digest. This is useful in addition to an immutable Hub revision when reproducing published results.

• **hub_kwargs – Extra arguments for huggingface_hub.hf_hub_download().

Returns:

A MicroDuckSkills object containing the frozen policy, ordered task library and action scale.