World age issue wtih read_pytorch

[Dsantra92]

read_pytorch throws world age issues.

julia> MLDatasets.read_pytorch("train.pt")
ERROR: MethodError: no method matching THload(::String)
The applicable method may be too new: running in world age 32451, while current world is 32460.
Closest candidates are:
  THload(::AbstractString) at ~/.julia/packages/Pickle/pwvBM/src/torch/torch_load.jl:62 (method too new to be called from this world context.)
  THload(::Pickle.Torch.TorchPickler, ::Any) at ~/.julia/packages/Pickle/pwvBM/src/torch/torch_load.jl:66 (method too new to be called from this world context.)
Stacktrace:
 [1] read_pytorch(path::String)
   @ MLDatasets ~/MLDatasets.jl/src/io.jl:28
 [2] top-level scope
   @ REPL[2]:1