Add JuliaRL_DQN_CartPole (#650)

* add back common networks

* add TwinNetwork

* sync

* add experiment JuliaRL_DQN_CartPole

Author: findmyway

Project.toml

@@ -12,10 +12,6 @@ ReinforcementLearningZoo = "d607f57d-ee1e-4ba7-bcf2-7734c1e31854"
 
 [compat]
 Reexport = "0.2, 1"
-ReinforcementLearningBase = "0.10"
-ReinforcementLearningCore = "0.9"
-ReinforcementLearningEnvironments = "0.7"
-ReinforcementLearningZoo = "0.6"
 julia = "1.6"
 
 [extras]

src/ReinforcementLearningCore/Project.toml

@@ -18,6 +18,7 @@ Reexport = "189a3867-3050-52da-a836-e630ba90ab69"
 ReinforcementLearningBase = "e575027e-6cd6-5018-9292-cdc6200d2b44"
 ReinforcementLearningEnvironments = "25e41dd2-4622-11e9-1641-f1adca772921"
 ReinforcementLearningTrajectories = "6486599b-a3cd-4e92-a99a-2cea90cc8c3c"
+Setfield = "efcf1570-3423-57d1-acb7-fd33fddbac46"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
 StatsBase = "2913bbd2-ae8a-5f71-8c99-4fb6c76f3a91"
 UnicodePlots = "b8865327-cd53-5732-bb35-84acbb429228"

src/ReinforcementLearningCore/src/policies/learners.jl

@@ -5,14 +5,14 @@ import Functors
 
 abstract type AbstractLearner end
 
-(L::AbstractLearner)(env) = env |> state |> send_to_device(L) |> L |> send_to_device(env)
+(L::AbstractLearner)(env::AbstractEnv) = env |> state |> send_to_device(L) |> L |> send_to_device(env)
 
 Base.@kwdef mutable struct Approximator{M,O}
     model::M
     optimiser::O
 end
 
-Functors.functor(x::Approximator) = (model = x.model,), y -> Approximator(y.model, x.state)
+Functors.functor(x::Approximator) = (model=x.model,), y -> Approximator(y.model, x.state)
 
 (A::Approximator)(x) = A.model(x)