Tutorial requires explicit gradient computation using Optimisers.jl, which does not align with new Zygote implicit gradient calls

[sampolk]

Flux 0.16 implemented some major changes that don't appear to be reflected in the main tutorial on using Transformers.jl (./docs/tutorial.md). Line 208 uses an implicit gradient call, but implicit gradients are no longer supported by Optimisers.jl since the Flux 0.16 update (link 1, link 2).