Hello!
I have been doing some deep reinforcement learning experiments with reinforcementlearning.jl and flux.jl in VScode windows. I have noticed that the first experiment i run after opening Vscode or after restarting the REPL is significantly faster than if i were to run the exact experiment a second time without restarting the REPL. Has anyone experienced this?
The difference between the exact sme script (experiment) is 5 minutes to 25.
Thanks, Jared