Add TotalRewardPerEpisodeLastN hook

src/ReinforcementLearningFarm/src/hooks/hooks.jl

@@ -1 +1 @@
-include("total_reward_per_episode_last_n.jl")
+include("total_reward_per_last_n_episodes.jl")

src/ReinforcementLearningFarm/src/hooks/total_reward_per_episode_last_n.jl → src/ReinforcementLearningFarm/src/hooks/total_reward_per_last_n_episodes.jl

@@ -5,47 +5,41 @@
 using CircularArrayBuffers: CircularVectorBuffer
 
 """
-    TotalRewardPerEpisodeLastN{F}(; max_steps = 100)
+TotalRewardPerLastNEpisodes{F}(; max_episodes = 100)
 
 A hook that keeps track of the total reward per episode for the last `max_steps` episodes.
 """
-struct TotalRewardPerEpisodeLastN{F} <: AbstractHook where {F<:AbstractFloat}
+struct TotalRewardPerLastNEpisodes{F} <: AbstractHook where {F<:AbstractFloat}
     rewards::CircularVectorBuffer{F}
 
-    function TotalRewardPerEpisodeLastN(; max_steps = 100)
+    function TotalRewardPerLastNEpisodes(; max_steps = 100)
         new{Float64}(CircularVectorBuffer{Float64}(max_steps))
     end
 end
 
-Base.getindex(h::TotalRewardPerEpisodeLastN{F}, inds...) where {F<:AbstractFloat} =
+Base.getindex(h::TotalRewardPerLastNEpisodes{F}, inds...) where {F<:AbstractFloat} =
     getindex(h.rewards, inds...)
 
 Base.push!(
-    h::TotalRewardPerEpisodeLastN{F},
+    h::TotalRewardPerLastNEpisodes{F},
     ::PostActStage,
     agent::P,
     env::E,
     player::Symbol,
 ) where {P<:AbstractPolicy,E<:AbstractEnv,F<:AbstractFloat} =
     h.rewards[end] += reward(env, player)
 
-function Base.push!(
-    hook::TotalRewardPerEpisodeLastN{F},
+Base.push!(
+    hook::TotalRewardPerLastNEpisodes{F},
     ::PreEpisodeStage,
     agent,
     env,
-) where {F<:AbstractFloat}
-    Base.push!(hook.rewards, 0.0)
-    return
-end
+) where {F<:AbstractFloat} = Base.push!(hook.rewards, 0.0)
 
-function Base.push!(
-    hook::TotalRewardPerEpisodeLastN{F},
+Base.push!(
+    hook::TotalRewardPerLastNEpisodes{F},
     stage::Union{PreEpisodeStage,PostEpisodeStage,PostExperimentStage},
     agent,
     env,
     player::Symbol,
-) where {F<:AbstractFloat}
-    Base.push!(hook, stage, agent, env)
-    return
-end
+) where {F<:AbstractFloat} = Base.push!(hook, stage, agent, env)

src/ReinforcementLearningFarm/test/hooks/hooks.jl

@@ -1 +1 @@
-include("total_reward_per_episode_last_n.jl")
+include("total_reward_per_last_n_episodes.jl")

src/ReinforcementLearningFarm/test/hooks/total_reward_per_episode_last_n.jl → src/ReinforcementLearningFarm/test/hooks/total_reward_per_last_n_episodes.jl

@@ -1,8 +1,8 @@
-using ReinforcementLearningFarm: TotalRewardPerEpisodeLastN
+using ReinforcementLearningFarm: TotalRewardPerLastNEpisodes
 
-@testset "TotalRewardPerEpisodeLastN" begin
+@testset "TotalRewardPerLastNEpisodes" begin
     @testset "Single Agent" begin
-        hook = TotalRewardPerEpisodeLastN(max_steps = 10)
+        hook = TotalRewardPerLastNEpisodes(max_steps = 10)
         env = TicTacToeEnv()
         agent = RandomPolicy()
 
@@ -15,7 +15,7 @@ using ReinforcementLearningFarm: TotalRewardPerEpisodeLastN
     end
 
     @testset "MultiAgent" begin
-        hook = TotalRewardPerEpisodeLastN(max_steps = 10)
+        hook = TotalRewardPerLastNEpisodes(max_steps = 10)
         env = TicTacToeEnv()
         agent = RandomPolicy()