你好，请问加载预训练模型spiking_resnet18，只能选择neuron.IFNode吗，可不可以使用neuron.ParametricLIFNode？

[hongliang520]

你好，请问加载预训练模型spiking_resnet18，只能选择neuron.IFNode吗，可不可以使用neuron.ParametricLIFNode？ resnet = spiking_resnet.spiking_wide_resnet50_2(pretrained=True, spiking_neuron=neuron.ParametricLIFNode, surrogate_function=surrogate.ATan(), detach_reset=True)

更改spiking_neuron=neuron.ParametricLIFNode会报错: RuntimeError: Error(s) in loading state_dict for SpikingResNet: Missing key(s) in state_dict: "sn1.w", "layer1.0.sn1.w", "layer1.0.sn2.w", "layer1.1.sn1.w", "layer1.1.sn2.w", "layer1.2.sn1.w", "layer1.2.sn2.w", "layer2.0.sn1.w", "layer2.0.sn2.w", "layer2.1.sn1.w", "layer2.1.sn2.w", "layer2.2.sn1.w", "layer2.2.sn2.w", "layer2.3.sn1.w", "layer2.3.sn2.w", "layer3.0.sn1.w", "layer3.0.sn2.w", "layer3.1.sn1.w", "layer3.1.sn2.w", "layer3.2.sn1.w", "layer3.2.sn2.w", "layer3.3.sn1.w", "layer3.3.sn2.w", "layer3.4.sn1.w", "layer3.4.sn2.w", "layer3.5.sn1.w", "layer3.5.sn2.w", "layer4.0.sn1.w", "layer4.0.sn2.w", "layer4.1.sn1.w", "layer4.1.sn2.w", "layer4.2.sn1.w", "layer4.2.sn2.w".

麻烦您有时间指点一下，谢谢

[fangwei123456]

ParametricLIFNode自带参数，pytorch提供的模型参数里面是缺失这一部分的。加载state dict的时候设置strict为false即可，具体请参考load state dict的文档

[hongliang520]

谢谢指点！