pat-coady trpo issues - Githubissues

pat-coady / trpo

Trust Region Policy Optimization with TensorFlow and OpenAI Gym

https://learningai.io/projects/2017/07/28/ai-gym-workout.html

MIT License

360 stars 106 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Can the `Data cardinality is ambiguous error` in Tensorflow 2.4 or 2.5 be solved as follows?

#39 wezardlza opened 3 years ago
1
is it TRPO?

#38 dim2r opened 3 years ago
0
Change of the loss function

#37 bernardocortez opened 4 years ago
0
Ocasional NaN's

#36 bernardocortez opened 4 years ago
0
Using train_on_batch

#35 ryanmaxwell96 opened 4 years ago
1
Saving and Loading trpo model for policynn

#34 ryanmaxwell96 opened 4 years ago
1
replace CartPoleBulletEnv-v1 by CartPoleContinuousBulletEnv-v0

#33 erwincoumans opened 4 years ago
0
Help Getting Cart Pole to Run

#32 ryanmaxwell96 closed 4 years ago
2
Graphing NN Model

#31 ryanmaxwell96 opened 4 years ago
0
Help understanding how to read the code

#30 ryanmaxwell96 opened 4 years ago
6
System Reboots

#29 vatsalkshah closed 5 years ago
0
able to run FetchPickAndPlace-v1 ?

#28 MrDadaGuy closed 4 years ago
2
Unusual replay buffer

#27 ghost closed 4 years ago
3
Temporal difference error in value estimates not calculated.

#26 ghost closed 5 years ago
0
Mistake in KL divergence formula

#25 ghost closed 5 years ago
0
Does this code use TRPO?

#24 ghost closed 5 years ago
1
Update train.py to avoid attribute errors

#23 gvgramazio closed 6 years ago
2
Fixes and code improvement on the code to save and restore trained models

#22 sanjaythakur closed 4 years ago
7
KL, PolicyEntropy, PolicyLoss go to NaN after 31,455 episodes

#21 David-Clement-Senbionic closed 6 years ago
6
error

#20 gautam1858 closed 6 years ago
2
Revert "Features to save and reuse the trained models are now integrated"

#19 pat-coady closed 6 years ago
0
Revert "Add PPO with clipping objective"

#18 pat-coady closed 6 years ago
0
Features to save and reuse the trained models are now integrated

#17 sanjaythakur closed 6 years ago
4
DOI for citation

#16 mkoseoglu closed 6 years ago
3
Add PPO with clipping objective

#15 magnusja closed 6 years ago
14
Trouble using pybullet and roboschool envs

#14 llecam closed 6 years ago
11
Why self.first_pass = True in Scaler method

#13 haoliuhl closed 6 years ago
1
enjoy a pre-trained model after training is done?

#12 erwincoumans closed 6 years ago
4
Rendering doesn't work. Window goes block

#11 abhishek2197 closed 6 years ago
4
Scaler vs. BatchNorm

#10 pender closed 7 years ago
3
training issue

#9 wenyijiang closed 6 years ago
4
Can't work on CartPole-v1

#8 AlexZhou1995 closed 7 years ago
1
CalledProcessError: Command '['avconv', '-version']' returned non-zero exit status 1

#7 FishQian closed 7 years ago
3
add command line arguments for network sizing and initial policy variance

#6 pat-coady closed 7 years ago
1
Some questions about the code

#5 20chase closed 7 years ago
1
Is there information on what actions and observations really are?

#4 wenyijiang closed 7 years ago
1
How much episodes should we do ?

#3 wenyijiang closed 7 years ago
1
Nice code! But much nicer if parallelized

#2 garymcintire closed 7 years ago
24
Roboschool issue (dimensionality of `action` in train.py:105)

#1 pender closed 7 years ago
2