search

glmcdona / LuxPythonEnvGym

Matching python environment code for Lux AI 2021 Kaggle competition, and a gym interface for RL models.
MIT License
73 stars 38 forks source link

Slow performance of map generation effecting training speed #77

Closed glmcdona closed 2 years ago

glmcdona commented 2 years ago

https://github.com/glmcdona/LuxPythonEnvGym/commit/f2dd9d3bd72a4f3a1c90a000fd4198c14a808f7f

Suggested solution is to use the built-in python random number generator if a seed isn't specified. That way it's much quicker because in training it won't be automatically generating 1M random numbers per map generation.