LoadError: UndefVarError: #flatten not defined

[smart-fr]

I think I messed with my installation of AlphaZero, and I can't recollect exactly how this occurred.

When launching a training, I get the following error:

Loading environment from: sessions/bonbon-rectangle-128_train

ERROR: LoadError: UndefVarError: #flatten not defined
Stacktrace:
  [1] deserialize_datatype(s::Serialization.Serializer{IOStream}, full::Bool)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:1364
  [2] handle_deserialize(s::Serialization.Serializer{IOStream}, b::Int32)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:866
  [3] deserialize(s::Serialization.Serializer{IOStream})
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:813
  [4] deserialize_datatype(s::Serialization.Serializer{IOStream}, full::Bool)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:1383
  [5] handle_deserialize(s::Serialization.Serializer{IOStream}, b::Int32)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:866
  [6] deserialize(s::Serialization.Serializer{IOStream})
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:813
  [7] deserialize_datatype(s::Serialization.Serializer{IOStream}, full::Bool)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:1388
  [8] handle_deserialize(s::Serialization.Serializer{IOStream}, b::Int32)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:866
  [9] deserialize(s::Serialization.Serializer{IOStream})
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:813
 [10] handle_deserialize(s::Serialization.Serializer{IOStream}, b::Int32)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:873
 [11] deserialize(s::Serialization.Serializer{IOStream}, t::DataType)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:1467
 [12] handle_deserialize(s::Serialization.Serializer{IOStream}, b::Int32)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:882
 [13] deserialize(s::Serialization.Serializer{IOStream})
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:813
 [14] handle_deserialize(s::Serialization.Serializer{IOStream}, b::Int32)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:919
 [15] deserialize
    @ C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:813 [inlined]
 [16] deserialize(s::IOStream)
    @ Serialization C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:800
 [17] open(f::typeof(Serialization.deserialize), args::String; kwargs::Base.Pairs{Symbol, Union{}, Tuple{}, NamedTuple{(), Tuple{}}})
    @ Base .\io.jl:384
 [18] open
    @ .\io.jl:381 [inlined]
 [19] deserialize
    @ C:\Program Files\Julia-1.8.5\share\julia\stdlib\v1.8\Serialization\src\Serialization.jl:810 [inlined]
 [20] load_env(dir::String)
    @ AlphaZero.UserInterface C:\Projets\BonbonRectangle\IA\dev\AlphaZero.jl\src\ui\session.jl:113
 [21] AlphaZero.UserInterface.Session(e::AlphaZero.Experiments.Experiment; dir::String, autosave::Bool, nostdout::Bool, save_intermediate::Bool)
    @ AlphaZero.UserInterface C:\Projets\BonbonRectangle\IA\dev\AlphaZero.jl\src\ui\session.jl:286
 [22] train(e::AlphaZero.Experiments.Experiment; args::Base.Pairs{Symbol, Any, Tuple{Symbol, Symbol}, NamedTuple{(:save_intermediate, :dir), Tuple{Bool, String}}})
    @ AlphaZero.Scripts C:\Projets\BonbonRectangle\IA\dev\AlphaZero.jl\src\scripts\scripts.jl:26
 [23] #train#15
    @ C:\Projets\BonbonRectangle\IA\dev\AlphaZero.jl\src\scripts\scripts.jl:28 [inlined]
 [24] top-level scope
    @ C:\Projets\BonbonRectangle\IA\dev\AlphaZero.jl\bin\main.jl:23
in expression starting at C:\Projets\BonbonRectangle\IA\dev\AlphaZero.jl\bin\main.jl:23

If I Google UndefVarError: #flatten not defined, I get in the release notes of the Flux.jl package that its recent version 0.13.14 fixed this issue: https://github.com/FluxML/Flux.jl/releases.

However, this version seems to be incompatible with AlphaZero.jl. Indeed, if I modify the file Project.toml in my customized version of AlphaZero.jl in order to update the Flux package to the 0.13.14 version, I get the following error:

ERROR: Unsatisfiable requirements detected for package cuDNN [02a925ec]:
 cuDNN [02a925ec] log:
 ├─possible versions are: 1.0.0-1.0.2 or uninstalled
 └─restricted by compatibility requirements with CUDA [052768ef] to versions: uninstalled — no versions left
   └─CUDA [052768ef] log:
     ├─possible versions are: 0.1.0-4.2.0 or uninstalled
     └─restricted to versions 3 by AlphaZero [7a1cc850], leaving only versions 3.0.0-3.13.1
       └─AlphaZero [7a1cc850] log:
         ├─possible versions are: 0.5.4 or uninstalled
         └─AlphaZero [7a1cc850] is fixed to version 0.5.4

Maybe because that version of Flux requires a version of CUDA which is incompatible with AlphaZero restrictions?

I don't know how to move forward.