hongshaorou/tdmpc2
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

update readme + dockerfile

Browse Source
This commit is contained in:
Nicklas Hansen
2025-05-13 14:51:16 -07:00
parent 7ec6bc83a8
commit 7992fa193e
2 changed files with 12 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
README.md
View File

@@ -36,19 +36,18 @@ You will need a machine with a GPU and at least 12 GB of RAM for single-task onl
We provide a `Dockerfile` for easy installation. You can build the docker image by running
```
cd docker && docker build . -t <user>/tdmpc2:1.0.0
cd docker && docker build . -t <user>/tdmpc2:1.0.1
```
This docker image contains all dependencies needed for running DMControl, Meta-World, and ManiSkill2 experiments.
This docker image contains all dependencies needed for running DMControl. We also provide a pre-built docker image [here](https://hub.docker.com/repository/docker/nicklashansen/tdmpc2/tags/1.0.1/sha256-b07d4e04d4b28ffd9a63ac18ec1541950e874bb51d276c7d09b36135f170dd93).
If you prefer to install dependencies manually, start by installing dependencies via `conda` by running the following command:
If you prefer to use `conda` rather than docker, start by running the following command:
```
conda env create -f docker/environment.yaml
pip install gym==0.21.0
```
The `environment.yaml` file installs dependencies required for training on DMControl tasks. Other domains can be installed by following the instructions in `environment.yaml`.
The `docker/environment.yaml` file installs dependencies required for training on DMControl tasks. Other domains can be installed by following the instructions in `docker/environment.yaml`.
If you want to run ManiSkill2, you will additionally need to download and link the necessary assets by running
@@ -62,13 +61,13 @@ which downloads assets to `./data`. You may move these assets to any location. T
export MS2_ASSET_DIR=<path>/<to>/<data>
```
and restart your terminal. Meta-World additionally requires MuJoCo 2.1.0. We host the unrestricted MuJoCo 2.1.0 license (courtesy of Google DeepMind) at [https://www.tdmpc2.com/files/mjkey.txt](https://www.tdmpc2.com/files/mjkey.txt). You can download the license by running
and restart your terminal. Note that Meta-World requires MuJoCo 2.1.0 and `gym==0.21.0` which is becoming increasingly difficult to install. We host the unrestricted MuJoCo 2.1.0 license (courtesy of Google DeepMind) at [https://www.tdmpc2.com/files/mjkey.txt](https://www.tdmpc2.com/files/mjkey.txt). You can download the license by running
```
wget https://www.tdmpc2.com/files/mjkey.txt -O ~/.mujoco/mjkey.txt
```
See `docker/Dockerfile` for installation instructions if you do not already have MuJoCo 2.1.0 installed. MyoSuite requires `gym==0.13.0` which is incompatible with Meta-World and ManiSkill2. Install separately with `pip install myosuite` if desired. Depending on your existing system packages, you may need to install other dependencies. See `docker/Dockerfile` for a list of recommended system packages.
Depending on your existing system packages, you may need to install other dependencies. See `docker/Dockerfile` for a list of recommended system packages.
----
@@ -87,7 +86,7 @@ This codebase provides support for all **104** continuous control tasks from **D
| myosuite | myo-key-turn
| myosuite | myo-key-turn-hard
which can be run by specifying the `task` argument for `evaluation.py`. Multi-task training and evaluation is specified by setting `task=mt80` or `task=mt30` for the 80-task and 30-task sets, respectively. We also provide basic support for other MuJoCo/Box2d Gymnasium tasks; refer to the `envs` directory for a list of tasks. It should be relatively straightforward to add support for custom tasks by following the examples in `envs`.
which can be run by specifying the `task` argument for `evaluation.py`. Multi-task training and evaluation is specified by setting `task=mt80` or `task=mt30` for the 80-task and 30-task sets, respectively. While you generally do not need to access the underlying task IDs or embeddings during training or evaluation of our multi-task models, the mapping from task name to task embedding used in our work can be found [here](https://github.com/nicklashansen/tdmpc2/blob/7ec6bc83a82a5188ca3faddc59aea83f430ab570/tdmpc2/common/__init__.py#L26). As of April 2025, our codebase also provides basic support for other MuJoCo/Box2d Gymnasium tasks; refer to the `envs` directory for a list of tasks. It should be relatively straightforward to add support for custom tasks by following the examples in `envs`.
**Note:** we also provide support for image observations in the DMControl tasks. Use argument `obs=rgb` if you wish to train visual policies.

36
docker/Dockerfile
View File

@@ -4,14 +4,14 @@
# https://www.tdmpc2.com #
# -------------------------------------- #
# Build instructions: #
# docker build . -t <user>/tdmpc2:1.0.0 #
# docker push <user>/tdmpc2:1.0.0 #
# docker build . -t <user>/tdmpc2:1.0.1 #
# docker push <user>/tdmpc2:1.0.1 #
# -------------------------------------- #
# Run: #
# docker run -i \ #
# -v <path>/<to>/tdmpc2:/tdmpc2 \ #
# --gpus all \ #
# -t <user>/tdmpc2:1.0.0 \ #
# -t <user>/tdmpc2:1.0.1 \ #
# /bin/bash #
##########################################
@@ -42,34 +42,8 @@ RUN conda update conda && \
conda init
SHELL ["/bin/bash", "-c"]
RUN echo "cd /root" >> /root/.bashrc
# mujoco 2.1.0
ENV MUJOCO_GL egl
ENV LD_LIBRARY_PATH /root/.mujoco/mujoco210/bin:${LD_LIBRARY_PATH}
RUN mkdir -p /root/.mujoco && \
wget https://www.tdmpc2.com/files/mjkey.txt && \
wget https://github.com/deepmind/mujoco/releases/download/2.1.0/mujoco210-linux-x86_64.tar.gz && \
tar -xzf mujoco210-linux-x86_64.tar.gz && \
rm mujoco210-linux-x86_64.tar.gz && \
mv mujoco210 /root/.mujoco/mujoco210 && \
mv mjkey.txt /root/.mujoco/mjkey.txt && \
find /root/.mujoco -uid 421709 -exec chown root:root {} \; && \
python -c "import mujoco_py"
# gym
RUN pip install gym==0.21.0
# metaworld
RUN pip install git+https://github.com/Farama-Foundation/Metaworld.git@04be337a12305e393c0caf0cbf5ec7755c7c8feb
# maniskill2
ENV MS2_ASSET_DIR /root/data
RUN pip install mani-skill2==0.4.1 && \
cd /root && \
python -m mani_skill2.utils.download_asset all -y
# myosuite (conflicts with meta-world / mani-skill2)
# RUN pip install myosuite
# image does not include metaworld, maniskill, myosuite
# these can be installed separately; see environment.yaml for details
# success!
RUN echo "Successfully built TD-MPC2 Docker image!"