https://github.com/dai-dao/PPO-Gluon

Implementation of PPO in Gluon / MXnet
https://github.com/dai-dao/PPO-Gluon

Last synced: 2 months ago
JSON representation

Implementation of PPO in Gluon / MXnet

• Host: GitHub
• URL: https://github.com/dai-dao/PPO-Gluon
• Owner: dai-dao
• Created: 2017-11-19T23:51:04.000Z (about 7 years ago)
• Default Branch: master
• Last Pushed: 2017-11-22T22:05:32.000Z (about 7 years ago)
• Last Synced: 2024-08-01T22:42:07.829Z (5 months ago)
• Language: Python
• Size: 374 KB
• Stars: 6
• Watchers: 2
• Forks: 4
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• Awesome-MXNet - PPO(Gluon:star:)

README

        
2. Add entropy term to encourage exploration

3. GAE

4. Distributional 

5. Other environments 

6. Bigger -> SLower nets

7. The exploration noise causes NAN gradients, thus NAN outputs

8. Need experience replay because it's OBVIOUSLY forgetting stuff from the past. 

9. Use OpenAI examples

1. Combine 2 nets into one -> Works -> Learns a bit slower I think

2. Tuned hyper-parameters, specifically the size of roll-outs, number of updates and batch size

3. Next step -> Try GAE estimation

4. After -> Train in distributed setting with harder environments

5. Compare to OpenAI baseline

6. Incorporate into StarCraft