search

miroesli / psscscs

Reinforcement Learning Battlesnake
MIT License
2 stars 1 forks source link

Improve training efficiency #25

Open Krocodial opened 4 years ago

Krocodial commented 4 years ago

In order to drastically speed up training times we should eventually artificially increase our training dataset sizes by reflecting the game board, this essentially halves the possible states we need to analyze. See here for an explanation https://nanonets.com/blog/data-augmentation-how-to-use-deep-learning-when-you-have-limited-data-part-2/ . If we include rotation in this mix we can cut possible states down to 1/8th of the original size

Swapnil-dx commented 4 years ago

I will try working on this

Swapnil-dx commented 4 years ago

In order to do this, the only way I see how to make changes to the agent.py file to include augmented frames. Should I bother doing this now or wait until we have a working MVP?

Krocodial commented 4 years ago

This is just an enhancement. Not needed for MVP, might be nice to improve training times since the competition is like 10 days away