fuse_fx#

classtorch.ao.quantization.quantize_fx.fuse_fx(model,fuse_custom_config=None,backend_config=None)[source]#

Fuse modules like conv+bn, conv+bn+relu etc, model must be in eval mode.Fusion rules are defined in torch.ao.quantization.fx.fusion_pattern.py

Parameters

• model (*) – a torch.nn.Module model

• fuse_custom_config (*) – custom configurations for fuse_fx.SeeFuseCustomConfig for more details

Return type

GraphModule

Example:

fromtorch.ao.quantizationimportfuse_fxm=Model().eval()m=fuse_fx(m)