weidler/rlaspa issues and pull requests

#45 - Updated the dqn policies

Pull Request - State: closed - Opened by adrigrillo almost 6 years ago

#44 - Optimize dqn

Pull Request - State: closed - Opened by adrigrillo almost 6 years ago

#43 - Make tensorboard logs get their own directories

Issue - State: closed - Opened by weidler almost 6 years ago

#42 - Use 4 frames of Race/Evasion as one state

Issue - State: open - Opened by alessandroscoppio almost 6 years ago

#41 - Multi task learning

Pull Request - State: closed - Opened by weidler almost 6 years ago - 1 comment

#40 - Add agent.save() and agent.load()

Issue - State: closed - Opened by dannigt almost 6 years ago

#39 - Double Check the Workings of all Networks

Issue - State: closed - Opened by weidler almost 6 years ago

#38 - Check how DQN memory influence training when using ckpts, eventually save memory in ckpts as well.

Issue - State: closed - Opened by alessandroscoppio almost 6 years ago

#37 - Develop a better metric for the progress report

Issue - State: closed - Opened by weidler almost 6 years ago - 5 comments

#36 - Have rendered testruns automatically saved to a gif file

Issue - State: closed - Opened by weidler almost 6 years ago

#35 - Make Race/Evasion tasks customizable (kwargs for the register)

Issue - State: closed - Opened by weidler almost 6 years ago

#34 - Create template for experiments in Wiki

Issue - State: closed - Opened by HansBambel almost 6 years ago

#33 - Prepare job script for Aachen cluster

Issue - State: closed - Opened by dannigt almost 6 years ago

#32 - Verify if GPU is indeed utilized

Issue - State: closed - Opened by dannigt almost 6 years ago

#31 - Exploration strategy

Issue - State: closed - Opened by adrigrillo almost 6 years ago

#30 - Allow the DQN Policy to backpropagate through the representation module

Issue - State: closed - Opened by weidler almost 6 years ago

#29 - Include MultiTask Possibility into Framework

Issue - State: closed - Opened by weidler almost 6 years ago

#28 - Create new Tasks

Issue - State: closed - Opened by HansBambel almost 6 years ago - 1 comment

#27 - Tensorize everything

Issue - State: closed - Opened by HansBambel almost 6 years ago

#26 - Batch Learning

Pull Request - State: closed - Opened by weidler almost 6 years ago

#25 - Tasks now inherit from gym.env and have an internal step counter.

Pull Request - State: closed - Opened by HansBambel almost 6 years ago - 2 comments

#24 - Check if input is as expected and throw meaningful errors.

Issue - State: open - Opened by weidler almost 6 years ago - 1 comment

#23 - Document Code, include type hinting!

Issue - State: open - Opened by weidler almost 6 years ago - 1 comment

#22 - Look at the heads of the Janus and Cerberus architecture

Issue - State: closed - Opened by HansBambel almost 6 years ago

#21 - Parallel approach integration

Pull Request - State: closed - Opened by weidler almost 6 years ago - 1 comment

#20 - Make tasks implement the gym interface

Issue - State: closed - Opened by weidler almost 6 years ago - 1 comment

#19 - Create statistics of training time and performance

Issue - State: closed - Opened by HansBambel almost 6 years ago - 2 comments

#18 - Being able to save and continue training a model

Issue - State: closed - Opened by HansBambel almost 6 years ago - 4 comments

#17 - Create different tasks/obstacle maps to "transfer"

Issue - State: closed - Opened by HansBambel almost 6 years ago - 3 comments

#16 - Batch Learning in Representation Learners

Issue - State: closed - Opened by weidler almost 6 years ago

#15 - Memory delay

Pull Request - State: closed - Opened by adrigrillo almost 6 years ago
Labels: enhancement

#14 - Memory delay

Pull Request - State: closed - Opened by adrigrillo almost 6 years ago
Labels: enhancement

#13 - Cleaned the repo, moved to playground folder

Pull Request - State: closed - Opened by adrigrillo almost 6 years ago

#12 - Solution to DDQN policies. Added counter and generic method to policy

Pull Request - State: closed - Opened by adrigrillo almost 6 years ago
Labels: bug, enhancement

#12 - Solution to DDQN policies. Added counter and generic method to policy

Pull Request - State: closed - Opened by adrigrillo almost 6 years ago
Labels: bug, enhancement

#11 - Framework adaptation

Pull Request - State: closed - Opened by adrigrillo almost 6 years ago - 2 comments
Labels: enhancement

#10 - Modular Framework for different RL Approaches

Issue - State: closed - Opened by weidler almost 6 years ago - 4 comments

#9 - Make pathing task use numpy arrays for faster conversions

Issue - State: closed - Opened by HansBambel almost 6 years ago - 3 comments

#8 - Implement

Issue - State: closed - Opened by weidler almost 6 years ago

#7 - Create DQN framework

Issue - State: closed - Opened by weidler almost 6 years ago - 2 comments
Labels: enhancement

#6 - Learn ObstaclePathing in latent space

Issue - State: closed - Opened by weidler almost 6 years ago

#5 - Implement SiameseAutoencoder

Issue - State: closed - Opened by weidler almost 6 years ago - 1 comment
Labels: enhancement

#4 - Try to learn ObstaclePathing

Issue - State: closed - Opened by weidler almost 6 years ago - 6 comments

#3 - Test RL agent on (padded) two-task latent representation

Issue - State: closed - Opened by dannigt almost 6 years ago

#2 - qwdqwd

Issue - State: closed - Opened by weidler almost 6 years ago

#1 - Try to learn Pathing in latent space

Issue - State: closed - Opened by weidler almost 6 years ago

GitHub / weidler/rlaspa issues and pull requests