lasseuth1 / blood_bowl2

Applying Reinforcement Learning to Blood Bowl
3 stars 4 forks source link

From log: The update variable is added twice. #3

Open greentec opened 5 years ago

greentec commented 5 years ago

From main.py,

https://github.com/lasseuth1/blood_bowl2/blob/dbe54bfca929ae411e7e70f9359fbbc8c899dc8a/rarity_of_events/main.py#L167

and I added print(update) for debugging after this line,

https://github.com/lasseuth1/blood_bowl2/blob/dbe54bfca929ae411e7e70f9359fbbc8c899dc8a/rarity_of_events/main.py#L182

and this prints out:

Updates 50, Episodes 4.0, Episodes this update 4.0, Total Timesteps 20, Reward 25.0, Mean Reward pr. Episode 6.25
49
Updates 150, Episodes 7.0, Episodes this update 3.0, Total Timesteps 40, Reward 30.0, Mean Reward pr. Episode 10.00
99
Updates 300, Episodes 12.0, Episodes this update 5.0, Total Timesteps 60, Reward 30.0, Mean Reward pr. Episode 6.00
149

The resume_update variable is larger than the update variable(50->150->300 is different with 49(50-1)->99(100-1)->149(150-1)). I think it's a bug, but if it's what you intended then I might be wrong, I'd appreciate it if you could let me know. Thank you.