hongshaorou/dreamerv3-torch
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

updated README

Browse Source
This commit is contained in:
NM512 2023-05-05 18:21:19 +09:00
parent d692b377ec
commit 3ebb8ad617
2 changed files with 9 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
README.md
View File

@ -1,8 +1,5 @@
# dreamerv3-torch # dreamerv3-torch
Pytorch implementation of [Mastering Diverse Domains through World Models](https://arxiv.org/abs/2301.04104v1). Pytorch implementation of [Mastering Diverse Domains through World Models](https://arxiv.org/abs/2301.04104v1). DreamerV3 is a scalable algorithm that outperforms previous approaches across various domains with fixed hyperparameters.
![1](https://user-images.githubusercontent.com/70328564/227377956-4a0d7e48-22fb-4f44-aa10-e5878a5ef901.png)
## Instructions ## Instructions
@ -16,23 +13,27 @@ python3 dreamer.py --configs defaults --task dmc_walker_walk --logdir ~/dreamerv
``` ```
Train the agent on Alien in Atari 100K: Train the agent on Alien in Atari 100K:
``` ```
python3 dreamer.py --configs defaults atari --task atari_alien --logdir ~/dreamerv3-torch/logdir/atari_alien python3 dreamer.py --configs defaults atari100k --task atari_alien --logdir ~/dreamerv3-torch/logdir/atari_alien
``` ```
Monitor results: Monitor results:
``` ```
tensorboard --logdir ~/dreamerv3-torch/logdir tensorboard --logdir ~/dreamerv3-torch/logdir
``` ```
## Evaluation Results
More results will be added in the future.
![dmc_vision](https://user-images.githubusercontent.com/70328564/236276650-ae706f29-4c14-4ed3-9b61-1829a1fdedae.png)
![atari100k](https://user-images.githubusercontent.com/70328564/236276669-16a56be3-40d6-49fd-befa-97c72b7d2460.png)
## ToDo ## ToDo
- [x] Prototyping - [x] Prototyping
- [x] Modify implementation details based on the author's implementation - [x] Modify implementation details based on the author's implementation
- [x] Evaluate on DMC vision - [x] Evaluate on DMC vision
- [ ] Evaluate on Atari 100K - [x] Evaluate on Atari 100K
- [ ] Add state input capability - [ ] Add state input capability
- [ ] Evaluate on DMC Proprio - [ ] Evaluate on DMC Proprio
- [ ] etc. - [ ] etc.
## Acknowledgments ## Acknowledgments
This code is heavily inspired by the following works: This code is heavily inspired by the following works:
- danijar's Dreamer-v3 jax implementation: https://github.com/danijar/dreamerv3 - danijar's Dreamer-v3 jax implementation: https://github.com/danijar/dreamerv3

3
configs.yaml
View File

@ -1,3 +1,4 @@
# defaults is for Vision DMC
defaults: defaults:
logdir: null logdir: null
@ -118,8 +119,6 @@ defaults:
disag_units: 400 disag_units: 400
disag_action_cond: False disag_action_cond: False
visual_dmc:
atari100k: atari100k:
steps: 4e5 steps: 4e5
action_repeat: 4 action_repeat: 4