Merge branch 'dev' into baselines-diffusion-policy

.gitignore

@@ -38,4 +38,4 @@ videos/
 
 # testing
 .coverage*
-htmlcov/
+htmlcov/

.pre-commit-config.yaml

@@ -0,0 +1,36 @@
+repos:
+  - repo: https://github.com/pre-commit/pre-commit-hooks
+    rev: v2.3.0
+    hooks:
+      - id: check-ast
+      - id: check-merge-conflict
+      - id: check-toml
+      # - id: check-yaml
+      - id: end-of-file-fixer
+        files: \.py$
+      - id: trailing-whitespace
+        files: \.py$
+  - repo: https://github.com/psf/black
+    rev: 22.10.0
+    hooks:
+      - id: black
+        exclude: 'warp_maniskill/.*|docs/.*|examples/.*'
+        args:
+          - --line-length=88
+  - repo: https://github.com/PyCQA/isort
+    rev: 5.12.0
+    hooks:
+      - id: isort
+        exclude: 'warp_maniskill/.*|docs/.*|examples/.*'
+        args:
+          - --profile=black
+  - repo: https://github.com/myint/autoflake
+    rev: v1.4
+    hooks:
+      - id: autoflake
+        exclude: 'warp_maniskill/.*|docs/.*|examples/.*'
+        args:
+          - -r
+          - --in-place
+          - --remove-unused-variables
+          # - --remove-all-unused-imports

CONTRIBUTING.md

@@ -16,17 +16,34 @@ pip install -e . # install MS2 locally
 pip install pytest coverage stable-baselines3 # add development dependencies for testing purposes
 ```
 
+Then to setup pre-commit, run
+
+```
+pre-commit install
+```
+
 ## Testing
 
 Testing is currently semi-automated and a WIP. We currently rely on coverage.py and pytest to test ManiSkill2.
 
 After you make changes, be sure to add any necessary tests to cover any new code in the `tests/` folder and run all the tests with the following command
 
 ```
-coverage run --source=mani_skill2/ -a -m pytest tests # run tests
-coverage html --include=mani_skill2/**/*.py # see the test coverage results
+pytest tests/ -m "not slow and gpu_sim"
+pytest tests/ -m "not slow and not gpu_sim"
 ```
 
+Note that we add a "not slow" argument which is to prevent testing on slow tests like download utility testing. There is also the "gpu_sim" pytest mark, which marks some tests as having to use the GPU simulation. These tests are separated as CPU simulation cannot run once GPU simulation has started and vice versa. With that in mind, any test that uses GPU simulation must add the `@pytest.mark.gpu_sim` decorator.
+
+<!-- ```
+coverage run --source=mani_skill2/ -a -m pytest tests -m "not slow" # run tests
+coverage html --include=mani_skill2/**/*.py # see the test coverage results
+``` -->
+<!-- 
+To skip generating a coverage report and also for easy debugging you can just run
+```
+pytest tests/ --pdb --pdbcls=IPython.terminal.debugger:Pdb -m "not slow"
+``` -->
 
 ## Building
 

README.md

@@ -30,7 +30,8 @@ Please refer to our [documentation](https://haosulab.github.io/ManiSkill2) to le
 From pip:
 
 ```bash
-pip install mani-skill2
+# This is temporary in order to install a in-dev version of sapien 3
+pip install manualtest/sapien-3.0.0.dev20240221+fa245b5-cp39-cp39-manylinux2014_x86_64.whl
 ```
 
 From github:

docs/source/concepts/observation.md

@@ -96,7 +96,7 @@ For `obs_mode="pointcloud"`:
 An "actor" is a fundamental object that represents a physical entity (rigid body) that can be simulated in SAPIEN (the backend of ManiSkill2). An articulated object is a collection of links interconnected by joints, and each link is also an actor. In SAPIEN, `scene.get_all_actors()` will return all the actors that are not links of articulated objects. The examples are the ground, the cube in [PickCube](./environments.md#pickcube-v0), and the YCB objects in [PickSingleYCB](./environments.md#picksingleycb-v0). `scene.get_all_articulations()` will return all the articulations. The examples are the robots, the cabinets in [OpenCabinetDoor](./environments.md#opencabinetdoor-v1), and the chairs in [PushChair](./environments.md#pushchair-v1). Below is an example of how to get actors and articulations in SAPIEN.
 
 ```python
-import sapien.core as sapien
+import sapien
 
 scene: sapien.Scene = ...
 actors = scene.get_all_actors()  # actors excluding links

examples/baselines/decision-transformer/.dockerignore

@@ -0,0 +1 @@
+decisiontransformer.ipynb

examples/baselines/decision-transformer/Dockerfile

@@ -0,0 +1,4 @@
+FROM ag2897387/maniskill2
+COPY requirements.txt .
+RUN pip install -r requirements.txt
+COPY . .

examples/baselines/decision-transformer/Readme.md

@@ -0,0 +1,32 @@
+# Decision Transformer
+
+This baseline is adapted from the `gym` example provided [here](https://github.com/kzl/decision-transformer/tree/master). This folder provides both a jupyter notebook version that can be run on Google Colab as well as a python script that can be used for training a model. A few environments have already been trained on and these runs can be viewed [here](https://wandb.ai/ag115115/decision-transformer).
+
+## Run Training Using Python Script
+
+### Using Docker
+
+1. Pull the docker image
+
+   `docker pull ag2897387/dec_tr_mani:latest`
+
+2. Run the training script
+
+   `docker run --rm -it --gpus all ag2897387/dec_tr_mani:latest python train.py --env LiftCube-v0 --dataset demos/v0/rigid_body/LiftCube-v0/trajectory.state.pd_ee_delta_pose.h5`
+
+There are additional arguments that specify parameters for the model and such.
+
+Results can also be logged to [Wandb](https://wandb.ai/) using the follow arguments.
+
+`docker run --rm -it --gpus all ag2897387/dec_tr_mani python train.py --dataset demos/v0/rigid_body/{env_id}/trajectory.state.pd_ee_delta_pose.h5 --log_to_wandb true --wandb_key <insert_wandb_key>`
+
+### Without Docker
+1. Install dependencies
+```
+pip install mani_skill2
+pip install -r requirements.txt
+```
+2. Run the training script
+```
+python train.py --env LiftCube-v0 --dataset demos/v0/rigid_body/LiftCube-v0/trajectory.state.pd_ee_delta_pose.h5
+```