ReinforcementLearning.jl

ReinforcementLearning.jl copied to clipboard

policy(env) returns no legal action -inf initialized Q-table

4

When initializing a Q-table with `-inf`, it looks that `EpsilonGreedyExplorer` might return no legal actions. MWE follows. Define a custom environment similar to RandomWalk1D ```julia Base.@kwdef mutable struct MyRandomWalk1D 1.0...

filchristou

Package Stabilization Plan

3

Hey! Right now it seems like the package is caught in the middle of a really important (but overwhelming refactor). Would the following approach be worth consideration? - [x] 1....

jeremiahpslewis

ReinforcementLearning.jl v0.12

22

# Components - [ ] https://github.com/JuliaReinforcementLearning/ReinforcementLearning.jl/issues/961 - [ ] https://github.com/JuliaReinforcementLearning/ReinforcementLearning.jl/issues/878 - [ ] https://github.com/JuliaReinforcementLearning/ReinforcementLearning.jl/issues/920 (?) - [ ] https://github.com/JuliaReinforcementLearning/ReinforcementLearning.jl/issues/954 (?)

jeremiahpslewis

It's not feasible to update the Q-based value agent in large steps for the RandomWalk1D() environment.

10

I followed the `RandomWalk1D()` example in the tutorial and wanted to update the agent. But `run` function returns `BoundsError: attempt to access 2×7 Matrix{Float64} at index [0, 1]` if I...

Van314159

Overspecific types in built-in policies and algorithms

2

The types in built-in policies and algorithms like `QBasedPolicy` and `TDLearner` are overly specific and prevent users from using the existing code to extend to new algorithms. Rather, it forces...

dharux

Update Website, fix broken references / issues

4

I've done an extensive first pass with the v0.11 release, but it looks like there's still some cleaning up to do, ~perhaps something @joelreymont you would be interested in tackling~...

jeremiahpslewis

Add note about RLZoo deprecation

3

PR Checklist - [ ] Update NEWS.md? - [ ] Unit tests for all structs / functions? - [ ] Integration and correctness tests using a simple env? - [...

jeremiahpslewis

See: https://github.com/JuliaReinforcementLearning/ReinforcementLearningAnIntroduction.jl

jeremiahpslewis

This pull request changes the compat entry for the `ReinforcementLearningTrajectories` package from `0.3.7` to `0.3.7, 0.4` for package ReinforcementLearningCore. This keeps the compat entries for earlier versions. Note: I have...

github-actions[bot]

This pull request changes the compat entry for the `GPUArrays` package from `8, 9, 10` to `8, 9, 10, 11` for package ReinforcementLearningCore. This keeps the compat entries for earlier...

github-actions[bot]