This is an example of Reinforcement learning running in Deno. Written on top of the tfjs.
Just a create (o
) trying to reach the goal (E
)
Nagranie.z.ekranu.2024-11-30.o.20.17.42.mov
https://docs.deno.com/runtime/getting_started/installation/
deno install
deno run train
And observe how this tiny creature is trying to reach the goal
- add environment variables for hyper parameters
- save and load models
- implement Replay Buffer for better performance
- add more scenarios: mazes, deadly obstacles, enemies, multiple agents