dqn.jl

function DQN_target(π, 𝒫, 𝒟, γ::Float32; kwargs...)
        𝒟[:r] .+ γ .* (1.f0 .- 𝒟[:done]) .* maximum(value(π, 𝒟[:sp]), dims=1)
end

function DQN(;π::DiscreteNetwork, 
              N::Int, 
              ΔN=4, 
              π_explore=ϵGreedyPolicy(LinearDecaySchedule(1., 0.1, floor(Int, N/2)), π.outputs), 
              c_opt::NamedTuple=(;), 
              log::NamedTuple=(;),
              c_loss=td_loss(),
              target_fn=DQN_target,
              prefix="",
              kwargs...)
              
     OffPolicySolver(;agent=PolicyParams(π=π, π_explore=π_explore, π⁻=deepcopy(π)), 
                      log=LoggerParams(;dir="log/dqn", log...),
                      N=N,
                      ΔN=ΔN,
                      c_opt = TrainingParams(;loss=c_loss, name=string(prefix, "critic_"), epochs=ΔN, c_opt...),
                      target_fn=target_fn,
                      kwargs...)
end