Official implementation of Direction-oriented Multi-objective Learning: Simple and Provable Stochastic Algorithms.

The expriments are conducted on Cityscapes and NYU-v2 datasets, which can be downloaded from MTAN. (Update: For Cityscapes, please choose the smaller version provided in the original MTAN repo). Following Nash-MTL and FAMO, the implementation is based on the MTL library.

Create the environment:

conda create -n mtl python=3.9.7
conda activate mtl
python -m pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113

Then, install the repo:

https://github.com/OptMN-Lab/sdmgrad.git
cd sdmgrad
python -m pip install -e .

The dataset by default should be put under experiments/EXP_NAME/dataset/ folder where EXP_NAME is chosen from nyuv2, cityscapes. To run the sdmgrad experiment:

cd experiments/EXP_NAME
sh run.sh

The experiments are conducted on Meta-World benchmark. To run the experiments on MT10 and MT50 (the instructions below are partly borrowed from CAGrad):

1. Create python3.6 virtual environment.
2. Install the MTRL codebase.
3. Install the Meta-World environment with commit id d9a75c451a15b0ba39d8b7a8b6d18d883b8655d8.
4. Copy the mtrl_files folder to the mtrl folder in the installed mtrl repo, then

cd PATH_TO_MTRL/mtrl_files/ && chmod +x mv.sh && ./mv.sh

1. Follow the run.sh to run the experiments.