dreamer_v3_12m_world_model(pretrained: bool = False, overrides: object = {})The smallest rung with all three objectives.
Model Size
Parameters
pretrainedbool= FalseNo weights are published for this family; passing
True raises.**overridesobject= {}Optional
DreamerV3Config field overrides.Returns
DreamerV3ForWorldModelingThe rung plus the world-model, actor and critic objectives.
Notes
Reference: Hafner, Pasukonis, Ba, and Norouzi, "Mastering Diverse Domains through World Models", Nature 640 (2025), 647-653 (arXiv:2301.04104).
The three losses take three separate optimisers; use
lucid.models.DreamerV3ForWorldModeling.backward to fill their
gradients, and call update_slow_critic once per step.
Examples
>>> from lucid.models import dreamer_v3_12m_world_model
>>> model = dreamer_v3_12m_world_model(action_dim=6).eval()
>>> model.config.free_nats, model.config.rep_scale
(1.0, 0.1)