Commit Graph

11 Commits

Author SHA1 Message Date
4b8ffeac6d GAIL V0.1 save point
GAIL V0.1 save point
todo 
1.human action record(GAILMem) debug
2.gail debug
2022-12-04 08:42:10 +09:00
ad9817e7a4 Totally disparate NN by target
Totally disparate NN by target.
2022-12-03 21:35:33 +09:00
cbc385ca10 Change training dataset storage method
save training dataset by it target type.
while training NN use single target training set to backward NN.
this improve at least 20 times faster than last update!
2022-12-03 07:54:38 +09:00
895cd5c118 Add EndReward Broadcast function
while game over add remaintime/15 to every step's rewards. to improve this round's training weight.
fix get target from states still using onehot decoder bug.
2022-12-03 03:58:19 +09:00
3930bcd953 Add Multi-NN agent
Add Multi neural network in output layer
use different nn while facing to different target.
2022-12-01 19:55:51 +09:00
5631569b31 Side Channel added
add side Channel to save target win ratio. 
Fix some Bug
2022-11-30 06:45:07 +09:00
32d398dbef Change Learning timing
change learning timing to each episode end.
2022-11-16 19:40:57 +09:00
a0895c7449 Add load & save function.
Add load & save function.
Add train flag to test model.
Add new action select function while in test mode.
Add decision period to skip step.
2022-11-08 23:14:34 +09:00
474032d1e8 hybrid dis-con action, save-load, converge wad observed
add discrete and continuous action in same NN model.
model save and load.
reward is increasing, converge was observed.

this two models are seems good:
Aimbot_9331_1667423213_hybrid_train2
Aimbot_9331_1667389873_hybrid
2022-11-03 07:16:18 +09:00
0dbe2013ae weight and bias sync added
weight and bias sync added
2022-11-01 19:11:45 +09:00
7497ffcb0f Parallel Environment Discrete PPO finish
Parallel Environment Discrete PPO finish. Runnable.
2022-10-30 04:13:14 +09:00