Skip to content

TD time step parameter #87

Description

@baedan

currently multi-step TD has an incorrect parameter (JuliaReinforcementLearning/ReinforcementLearning.jl#648).

function run_once(n, α)
env = StateTransformedEnv(
RandomWalk1D(N=NS, actions=ACTIONS),
state_mapping=GroupMapping(n=NS)
)
agent = Agent(
policy=VBasedPolicy(
learner=TDLearner(
approximator=TabularVApproximator(;
n_state=n_groups+2,
opt=Descent(α)
),
method=:SRS,
n=n
),
mapping=(env,V) -> rand(action_space(env))
),
trajectory=VectorSARTTrajectory()
)
hook = RecordRMS()
run(agent, env, StopAfterEpisode(10),hook)
mean(hook.rms)
end

as an example, the n is used as the number of time steps. however it currently corresponds to the number of time steps plus one. run_once(1, α) thus is not TD(0) which has a time step parameter of 1, but rather a 2-step TD method. depending on how upstream is resolved an update might be needed here.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions