popgym.baselines.models.indrnn

Module Contents

Classes

IndRNN

class popgym.baselines.models.indrnn.IndRNN(input_size, hidden_size, activation='relu', clamp=True, max_len=1024)

Bases: torch.nn.Module

clamp_weights() → None

forward(x: torch.Tensor, state: torch.Tensor | None = None) → Tuple[torch.Tensor, torch.Tensor]

input:

x: [B, T, F] state: [B, 2, F]

output:

y: [B, T, D] state: [B, 2, D]