protomotions.agents.supervised package

Supervised imitation and distillation agents.

The package owns the generic supervised rollout loop and named student presets such as MaskedMimic. Keep student architectures self-contained; share the loop and key-based supervision losses, not architecture-specific inheritance.

Submodules

• protomotions.agents.supervised.agent module
  • SupervisedAgent
    • model
    • create_model()
    • create_optimizers()
    • register_algorithm_experience_buffer_keys()
    • register_algorithm_experience_buffer_keys_from_obs()
    • collect_rollout_step()
    • perform_optimization_step()
    • supervised_step()
    • calculate_extra_loss()
    • get_state_dict()
• protomotions.agents.supervised.config module
  • RolloutActor
    • STUDENT
    • EXPERT
    • from_str()
  • SupervisedAgentConfig
    • model
    • expert_model_path
    • rollout_actor
    • loss
    • __init__()
• protomotions.agents.supervised.expert_utils module
  • get_expert_actor_in_keys()
  • get_expert_observation_keys()
  • get_expert_observation_components()
• protomotions.agents.supervised.masked_mimic_config module
  • KLDScheduleConfig
    • init_kld_coeff
    • end_kld_coeff
    • start_epoch
    • end_epoch
    • __init__()
  • VAENoiseType
    • NORMAL
    • UNIFORM
    • ZEROS
    • from_str()
  • MaskedMimicVAEConfig
    • kld_schedule
    • vae_latent_dim
    • vae_noise_type
    • __init__()
  • MaskedMimicModelConfig
    • encoder
    • prior
    • trunk
    • vae
    • optimizer
    • __init__()
  • MaskedMimicSupervisedAgentConfig
    • model
    • loss
    • __init__()
• protomotions.agents.supervised.masked_mimic_model module
  • MaskedMimicModel
    • __init__()
    • rollout_state_specs()
    • forward()
    • forward_inference()
    • get_inference_in_keys()
    • kl_loss()
    • compute_model_loss()