Library of Layers adaptively orders transformer layers at runtime

The Library of Layers paper presents a method for adaptively ordering transformer layers during runtime. The submitted work explores changing layer execution order without retraining the model.