yenchenlin / DeepLearningFlappyBird

Flappy Bird hack using Deep Reinforcement Learning (Deep Q-learning).
MIT License
6.62k stars 2.04k forks source link

FINAL_EPSILON = 0.0001 ,maybe INITIAL_EPSILON = 1? #40

Closed MemoryCrash closed 6 years ago

MemoryCrash commented 6 years ago
   FINAL_EPSILON = 0.0001 
   INITIAL_EPSILON = 0.0001 
   epsilon = INITIAL_EPSILON
    "so in this condition "epsilon" will never be update."
    if epsilon > FINAL_EPSILON and t > OBSERVE:
        epsilon -= (INITIAL_EPSILON - FINAL_EPSILON) / EXPLORE