ritchie
|
1b0ebe6683
|
goal proper added in her
|
2017-12-16 16:34:27 +01:00 |
|
ritchie
|
c7d6fea511
|
fixed reward function -> less deflection is rewarding
|
2017-12-16 15:26:44 +01:00 |
|
ritchie
|
599420dbe6
|
no double elements anymore in her_deep_Q_bridge.ipynb. deploy instead of last action
|
2017-12-16 15:12:27 +01:00 |
|
vik
|
8e6f17f036
|
log regression formula update
|
2017-11-15 13:26:45 +01:00 |
|
vik
|
c21a0681d0
|
popleft bug buffer fixed and double deep q learning added
|
2017-11-06 16:17:50 +01:00 |
|
vik
|
86341c51ab
|
more stable learning due to target network
|
2017-11-06 11:01:38 +01:00 |
|
Ritchie
|
7e7d931adc
|
Q algorithm learns
|
2017-11-04 13:17:22 +01:00 |
|
Ritchie
|
40dcf31329
|
policy network openai gym flagpole
|
2017-10-31 22:30:20 +01:00 |
|
Ritchie
|
5b68b7ec43
|
logistic regression
|
2017-09-10 13:54:17 +02:00 |
|
Ritchie
|
f351e89421
|
readme
|
2017-09-03 12:50:43 +02:00 |
|
Ritchie
|
7dfe24aa76
|
vanilla_mlp
|
2017-07-13 11:28:12 +02:00 |
|
Ritchie
|
36ca828e6b
|
No activation prime
|
2017-07-01 12:20:00 +02:00 |
|
Ritchie
|
43a8098031
|
learning for both nn working
|
2017-07-01 12:01:07 +02:00 |
|
Ritchie
|
ea28e2ee29
|
Backprop working
|
2017-06-30 18:13:59 +02:00 |
|
Ritchie
|
70b4cc7bc9
|
Backprop not exploding
|
2017-06-30 17:49:09 +02:00 |
|
Ritchie
|
5c56c91a4a
|
backprop though not working
|
2017-06-30 17:03:58 +02:00 |
|
Ritchie
|
a3c38d2ad7
|
Various layer network. Feed forward
|
2017-06-30 14:11:02 +02:00 |
|
Ritchie
|
c21622525c
|
cross entropy delta
|
2017-06-25 22:05:55 +02:00 |
|
Ritchie
|
add5381051
|
first
|
2017-06-24 21:13:41 +02:00 |
|