What is the best way to define efficientnet models in fastai!

I wonder, what is the easiest way to define the efficientNet models with fastai?

Hello,

fastai supports an excellent computer vision library, timm, that contains a panoply of image models with pre-trained parameters, including EfficientNet. timm’s models can be found here, and you simply pass the name of your desired architecture to fastai’s vision_learner.

For instance, the segment below creates a Learner around EfficientNet-B0.

learn = vision_learner(dls, 'efficientnet_b0')

Does that answer your question?

3 Likes

yes, this is what i need . Thanks

1 Like