AttributeError: 'SqueezeExcite' object has no attribute 'gate'

[isuranga1]

When I load the efficient net model it works fine with the timm version i used which is timm==0.4.12 , but when I use the forward function to input an image tensor, it gives me this error. Does anyone know how to resolve it?

[rwightman]

either a mismatch of timm component versions across your environment/local code, or model weights were serialized by pickling the whole model instead of just the state dict.

[isuranga1]

Thank you I tried downgrading the timm version and eventually it worked.