IDQLMlp

CLASS cleandiffuser.nn_diffusion.IDQLMlp(obs_dim: int, act_dim: int, emb_dim: int = 64, hidden_dim: int = 256, n_blocks: int = 3, dropout: float = 0.1, timestep_emb_type: str = “positional”, timestep_emb_params: Optional[dict] = None) [SOURCE]

A carefully designed MLP neural network backbone with redidual structure and layer normalization, proposed in Implicit Diffusion Q-Learning (IDQL). It takes the current observation as the context tensor and generates the action to execute.

Parameters:

• obs_dim (int): The dimension of the observation tensor \(\bm o_t\).
• act_dim (int): The dimension of the action tensor \(\bm x_t\).
• emb_dim (int): The dimension of the time embedding. Default is 64.
• hidden_dim (int): The dimension of the hidden layers of the MLP. Default is 256.
• n_blocks (int): The number of residual blocks. Default is 3.
• dropout (float): The dropout rate. Default is 0.1.
• timestep_emb_type (str): The type of the time embedding. It can be either “positional” or “fourier”. Default is “positional”.
• timestep_emb_params (Optional[dict]): The parameters for the time embedding. Default is None.

forward(x: torch.Tensor, t: torch.Tensor, c: torch.Tensor) -> torch.Tensor

Parameters:

• x (torch.Tensor): The input tensor \(\bm x_t\) in shape (..., act_dim)
• t (torch.Tensor): The time tensor \(t\) in shape (..., 1).
• c (torch.Tensor): The context tensor \(\bm c\) in shape (..., obs_dim).

Returns:

• torch.Tensor: The output tensor in shape (..., act_dim).