Skip to content

Latest commit

 

History

History
53 lines (46 loc) · 1.13 KB

README.md

File metadata and controls

53 lines (46 loc) · 1.13 KB

Context is Everything: Implicit Identification for Dynamics Adaptation

Project Page

Installation Instructions

  1. Clone the repo
git clone git@github.com:bennevans/iida.git
cd iida/

  1. Install MuJoCo here.

  2. Install the required python packages:

conda env create -f setup/environment.yaml

OR

cd setup/
pip install -r requirements.txt
  1. Install the varyingsim library
pip install -e .
  1. Generate the simulated dataset or download it here.

Dataset Generation Instructions

Single-step Environments

cd scripts/dataset/
python create_da_dataset_push_box.py
python create_da_dataset_slide_puck.py

Multi-step Environments

To train a policy to generate rollouts, I use mjrl.

cd scripts/policy/
python train_policy.py
python relabel_paths.py --env <env>

Run instructions

To run with the current experiment and algorithm parameters:

cd scripts/

And run:

python run_exp.py -c <config file>.yaml