Rate this Page

Conv1d#

class torch.ao.nn.qat.modules.conv.Conv1d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True, padding_mode='zeros', qconfig=None, device=None, dtype=None)[source]#

A Conv1d module attached with FakeQuantize modules for weight, used for quantization aware training.

We adopt the same interface as Conv1d

Similar to Conv2d, with FakeQuantize modules initialized to default.

Variables:

weight_fake_quant – fake quant module for weight