causal-conv1d 1.1.1

Causal depthwise conv1d in CUDA, with a PyTorch interface

Causal depthwise conv1d in CUDA with a PyTorch interface

Features:

• Support fp32, fp16, bf16.
• Kernel size 2, 3, 4.

How to use

from causal_conv1d import causal_conv1d_fn

def causal_conv1d_fn(x, weight, bias=None, activation=None):
    """
    x: (batch, dim, seqlen)
    weight: (dim, width)
    bias: (dim,)
    activation: either None or "silu" or "swish"

    out: (batch, dim, seqlen)
    """

Equivalent to:

import torch.nn.functional as F

F.conv1d(x, weight.unsqueeze(1), bias, padding=width - 1, groups=dim)[..., :seqlen]