Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/caiqi/Silver-Bullet-3D
This repository is the official implementation of *Silver-Bullet-3D* Solution for SAPIEN ManiSkill Challenge 2021
https://github.com/caiqi/Silver-Bullet-3D
Last synced: 3 days ago
JSON representation
This repository is the official implementation of *Silver-Bullet-3D* Solution for SAPIEN ManiSkill Challenge 2021
- Host: GitHub
- URL: https://github.com/caiqi/Silver-Bullet-3D
- Owner: caiqi
- Created: 2022-01-18T14:46:55.000Z (almost 3 years ago)
- Default Branch: master
- Last Pushed: 2022-01-19T09:36:46.000Z (almost 3 years ago)
- Last Synced: 2024-10-22T09:48:52.158Z (25 days ago)
- Language: Python
- Size: 186 KB
- Stars: 20
- Watchers: 1
- Forks: 2
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-decision-transformer - official
README
# Introduction
This repository is the official implementation of *Silver-Bullet-3D* Solution for SAPIEN ManiSkill Challenge 2021
## Requirement:
* PyTorch 1.8.0+
* Python3.7
* CUDA 10.1+
Other requirements please refer to [environment.yml](No_Interaction/evaluation/bucket_track1/environment.yml)
## Clone the repository:
```
git clone https://github.com/caiqi/Silver-Bullet-3D
```
## No Interaction Track
### Data preparation
* Download ManiSkill dataset from [here](https://github.com/haosulab/ManiSkill)
* Compress the data with [compress_data.sh](No_Interaction/training/scripts/compress_data.sh)
### Training
Training code is provided in [No_Interaction/training](No_Interaction/training) folder. For example, to train MoveBucket model, using following script:
```
CONFIG_NAME=bucket/1225_bucket_ensemble_v1
SEED=1345
CUDA_VISIBLE_DEVICES=0 python -m tools.run_rl configs/${CONFIG_NAME}.py --gpu-ids=0 --seed ${SEED} --work-dir ${CONFIG_NAME}
```
For final submission, we ensemble multiple models with different network architecture and random seed. All configs are provided in [configs](No_Interaction/training/configs) folder.
### Evaluation
Evaluation code and checkpoints are provided in [No_Interaction/evaluation](No_Interaction/evaluation). For example, to evaluate the pre-trained models on MoveBucket, use ManiSkill official evaluation code:
```
PYTHONPATH=No_Interaction/evaluation/bucket_track1:$PYTHONPATH python evaluate_policy.py --env MoveBucket-v0 --level-range "0-300"
```
[evaluate_policy.py](https://github.com/haosulab/ManiSkill/blob/main/mani_skill/tools/evaluate_policy.py) is from [ManiSkill](https://github.com/haosulab/ManiSkill) repo. Checkpoints can be downloaded from release page.
| Task | Models |
| :---------------: | :----------------------------------------------------------------------------------------------: |
| OpenCabinetDoor | [Checkpoint](https://github.com/caiqi/Silver-Bullet-3D/releases/download/v1.0/door_track1.zip) |
| OpenCabinetDrawer | [Checkpoint](https://github.com/caiqi/Silver-Bullet-3D/releases/download/v1.0/drawer_track1.zip) |
| MoveBucket | [Checkpoint](https://github.com/caiqi/Silver-Bullet-3D/releases/download/v1.0/bucket_track1.zip) |
| PushChair | [Checkpoint](https://github.com/caiqi/Silver-Bullet-3D/releases/download/v1.0/chair_track1.zip) |
## No Restriction Track
The training and evaluation code is the same.
| Task | Code |
| :---------------: | :---------------------------------------------------------------: |
| OpenCabinetDoor | [user_solution_door.py](No_Restriction/user_solution_door.py) |
| OpenCabinetDrawer | [user_solution_drawer.py](No_Restriction/user_solution_drawer.py) |
| MoveBucket | [user_solution_bucket.py](No_Restriction/user_solution_bucket.py) |
| PushChair | [user_solution_chair.py](No_Restriction/user_solution_chair.py) |
## Acknowledgements
Thanks the contribution of [ManiSkill-Learn](https://github.com/haosulab/ManiSkill-Learn) and awesome PyTorch team.