Fix mlp nonlinearity

[ziqi-ma]

Fix nonlinearity of final MLP - otherwise output is lower-bounded by GELU.

[JeanKossaifi]

Thanks @ziqi-ma! @m4e7 this fixes the issue you found too right?

[m4e7]

Thanks @ziqi-ma! @m4e7 this fixes the issue you found too right?

Yeah, this fixes it! We should also fix MLPLinear.forward(), too: https://github.com/neuraloperator/neuraloperator/blob/a371d0890698eb2617dc805b527449b7acac8581/neuralop/layers/mlp.py#L77C34-L77C34

[JeanKossaifi]

@ziqi-ma would you be able to make that change too?

[JeanKossaifi]

I'll merge this @ziqi-ma as this is an important fix and opened #200 to track the other issue if you want to open a PR for it too! Thanks for your contribution!