weidler RLaSpa issues - Githubissues

weidler / RLaSpa

Reinforcement Learning in Latent Space

MIT License

5 stars 1 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Updated the dqn policies

#45 adrigrillo closed 5 years ago
0
Optimize dqn

#44 adrigrillo closed 5 years ago
0
Make tensorboard logs get their own directories

#43 weidler closed 5 years ago
0
Use 4 frames of Race/Evasion as one state

#42 alessandroscoppio opened 5 years ago
0
Multi task learning

#41 weidler closed 5 years ago
1
Add agent.save() and agent.load()

#40 dannigt closed 5 years ago
0
Double Check the Workings of all Networks

#39 weidler closed 5 years ago
0
Check how DQN memory influence training when using ckpts, eventually save memory in ckpts as well.

#38 alessandroscoppio closed 5 years ago
0
Develop a better metric for the progress report

#37 weidler closed 5 years ago
5
Have rendered testruns automatically saved to a gif file

#36 weidler closed 5 years ago
0
Make Race/Evasion tasks customizable (kwargs for the register)

#35 weidler closed 5 years ago
0
Create template for experiments in Wiki

#34 HansBambel closed 5 years ago
0
Prepare job script for Aachen cluster

#33 dannigt closed 5 years ago
0
Verify if GPU is indeed utilized

#32 dannigt closed 5 years ago
0
Exploration strategy

#31 adrigrillo closed 5 years ago
0
Allow the DQN Policy to backpropagate through the representation module

#30 weidler closed 5 years ago
0
Include MultiTask Possibility into Framework

#29 weidler closed 5 years ago
0
Create new Tasks

#28 HansBambel closed 5 years ago
1
Tensorize everything

#27 HansBambel closed 5 years ago
0
Batch Learning

#26 weidler closed 5 years ago
0
Tasks now inherit from gym.env and have an internal step counter.

#25 HansBambel closed 5 years ago
2
Check if input is as expected and throw meaningful errors.

#24 weidler opened 5 years ago
1
Document Code, include type hinting!

#23 weidler opened 5 years ago
1
Look at the heads of the Janus and Cerberus architecture

#22 HansBambel closed 5 years ago
0
Parallel approach integration

#21 weidler closed 5 years ago
1
Make tasks implement the gym interface

#20 weidler closed 5 years ago
1
Create statistics of training time and performance

#19 HansBambel closed 5 years ago
2
Being able to save and continue training a model

#18 HansBambel closed 5 years ago
4
Create different tasks/obstacle maps to "transfer"

#17 HansBambel closed 5 years ago
3
Batch Learning in Representation Learners

#16 weidler closed 5 years ago
0
Memory delay

#15 adrigrillo closed 5 years ago
0
Memory delay

#14 adrigrillo closed 5 years ago
0
Cleaned the repo, moved to playground folder

#13 adrigrillo closed 5 years ago
0
Solution to DDQN policies. Added counter and generic method to policy

#12 adrigrillo closed 5 years ago
0
Framework adaptation

#11 adrigrillo closed 5 years ago
2
Modular Framework for different RL Approaches

#10 weidler closed 5 years ago
4
Make pathing task use numpy arrays for faster conversions

#9 HansBambel closed 5 years ago
3
Implement

#8 weidler closed 5 years ago
0
Create DQN framework

#7 weidler closed 5 years ago
2
Learn ObstaclePathing in latent space

#6 weidler closed 5 years ago
0
Implement SiameseAutoencoder

#5 weidler closed 5 years ago
1
Try to learn ObstaclePathing

#4 weidler closed 5 years ago
6
Test RL agent on (padded) two-task latent representation

#3 dannigt closed 5 years ago
0
qwdqwd

#2 weidler closed 5 years ago
0
Try to learn Pathing in latent space

#1 weidler closed 5 years ago
0