nn.utils.rnn.pad_sequence

lucid.nn.utils.rnn.pad_sequence(sequences: Iterable[Tensor], batch_first: bool = False, padding_value: int | float | complex = 0) → Tensor

Function Signature

def pad_sequence(
    sequences: Iterable[Tensor],
    batch_first: bool = False,
    padding_value: _Scalar = 0.0,
) -> Tensor

Parameters

  • sequences (Iterable[Tensor]): A list (or iterable) of sequence tensors. Each tensor must share the same trailing shape (all dimensions except the first), and all tensors must be on the same device with the same dtype. The first dimension is treated as the sequence length.

  • batch_first (bool, optional): If True, output shape is (batch, max_len, *trailing_shape). If False, output shape is (max_len, batch, *trailing_shape). Default is False.

  • padding_value (_Scalar, optional): Value used for padding shorter sequences. Default is 0.0.

Return Value

  • Tensor: A padded tensor containing all sequences stacked along a new batch axis.

Behavior

  • The maximum sequence length is computed from the input list.

  • Sequences shorter than max_len are padded with padding_value.

  • This function expects a non-empty iterable of tensors.

Examples

import lucid
import lucid.nn as nn

a = lucid.Tensor([[1.0, 2.0], [3.0, 4.0]])      # length 2
b = lucid.Tensor([[5.0, 6.0], [7.0, 8.0], [9.0, 10.0]])  # length 3

out = nn.utils.rnn.pad_sequence([a, b], batch_first=False, padding_value=0.0)
print(out.shape)  # (3, 2, 2)

Usage Tips

Tip

Use batch_first=True if your RNN modules expect input shaped as (batch, seq, feature).