JuliaReinforcementLearning / ReinforcementLearning.jl

A reinforcement learning package for Julia
https://juliareinforcementlearning.org
Other
582 stars 112 forks source link

experiments failed #982

Closed Yue-Wang-qvp closed 7 months ago

Yue-Wang-qvp commented 1 year ago

When I execute JuliaRL_DDPG_Pendulum.jl, It was thrown that ERROR: LoadError: UndefVarError: params not defined. My package status reads as follows:

[79e6a3ab] Adapt v3.6.2 ⌅ [4fba245c] ArrayInterface v6.0.25 [c52e3926] Atom v0.12.38 [6e4b80f9] BenchmarkTools v1.3.2 [6cde8614] BlockDecomposition v1.13.0 [a076750e] CPLEX v1.0.0 [336ed68f] CSV v0.10.11 [9961bab8] Cbc v1.2.0 [aaaa29a8] Clustering v0.15.4
[88b4ec78] Coluna v0.7.0 [861a8166] Combinatorics v1.0.2
[a81c6b42] Compose v0.9.5
[a6e380b2] ControlSystems v1.8.0 [e993076c] DASSL v2.6.1 [717857b8] DSP v0.7.9 ⌃ [a93c6f00] DataFrames v1.3.6
[864edb3b] DataStructures v0.18.15 ⌃ [0c46a032] DifferentialEquations v7.6.0 [31c24e10] Distributions v0.25.102 [191a621a] Dualization v0.5.8
[7a1cc6ca] FFTW v1.7.1 ⌅ [587475ba] Flux v0.13.4 [60bf3e95] GLPK v1.1.3 [a2cc645c] GraphPlot v0.5.2
[bd48cda9] GraphRecipes v0.5.12
[86223c79] Graphs v1.9.0 [2e9cd046] Gurobi v1.0.3 ⌅ [cd3eb016] HTTP v0.9.17 [87dc4568] HiGHS v1.6.1 [7869d1d1] IRTools v0.4.10
[2030c09a] InfrastructureModels v0.7.8 ⌅ [8197267c] IntervalSets v0.5.4
[b6b21f68] Ipopt v1.4.2 [682c06a0] JSON v0.21.4 [4076af6c] JuMP v1.15.1 [e5e0dc1b] Juno v0.8.4 [b4f0291d] LazySets v2.11.0
⌃ [bdcacae8] LoopVectorization v0.12.150 [88578114] MATDaemon v0.1.2
[10e44e05] MATLAB v0.8.4 [b8f27783] MathOptInterface v1.20.1 [442fdcdd] Measures v0.3.2
[f28f55f0] Memento v1.4.1
[bcdb8e00] Metaheuristics v3.3.3 [6405355b] Mosek v10.1.3 [1ec41992] MosekTools v0.15.1
[6f286f6a] MultivariateStats v0.10.2 [2774e3e8] NLsolve v4.5.1
[fdc4e326] ODEProblemLibrary v0.1.7 [429524aa] Optim v1.7.7 [bac558e1] OrderedCollections v1.6.2 [9b87118b] PackageCompiler v2.1.9 [a03496cd] PlotlyBase v0.8.19
[f0f68f2c] PlotlyJS v0.18.10
[f2990250] PlotlyKaleido v2.2.0
[91a5bcdd] Plots v1.39.0 [67491407] Polyhedra v0.7.6
[c36e90e8] PowerModels v0.19.9
[d7431456] PowerModelsDistribution v0.15.0 ⌃ [e690365d] PowerSimulations v0.15.7 ⌅ [bcd98974] PowerSystems v1.25.3
[6f49c342] RCall v0.13.18
[158674fc] ReinforcementLearning v0.10.2 [295af30f] Revise v3.5.6 [21fd56a4] RobustAndOptimalControl v0.4.28 [82193955] SCIP v0.11.14 [c946c3f1] SCS v1.3.1 [c35d69d1] SMTPClient v0.6.4
[860ef19b] StableRNGs v1.0.0
⌅ [2913bbd2] StatsBase v0.33.21
[f3b207a7] StatsPlots v0.15.6
⌃ [c3572dad] Sundials v4.15.1
[04da0e3b] TypeTree v0.3.0
[fdbf4ff8] XLSX v0.10.0 [d00139f3] METIS_jll v5.1.2+0 [ade2ca70] Dates [f43a241f] Downloads v1.6.0
[b77e0a4c] InteractiveUtils
[37e2e46d] LinearAlgebra [9a3f8284] Random [2f01184e] SparseArrays [10745b16] Statistics v1.9.0

halyusuf25 commented 1 year ago

i have the same problem and it turns that params() function has been removed from Flux version 0.13 and more

HenriDeh commented 1 year ago

Hello,

The params() issue is fixed on the master head, but many algorithms are broken there, so we cannot make a release. This is due to an ongoing refactor of the package that we lack resources to complete.

A solution to your problem could be to use a local copy of RL.jl and edit Flux's compatibility to be 0.12.