🤖 Teaching Embodied Reinforcement Learning Agents: Informativeness and Diversity of Language Use

⚒️ Environment Setup

HomeGrid

Create conda env.

conda create -n homegrid-ltdt python=3.8
conda activate homegrid-ltdt

Install packages.

cd <path/to/homegrid>
pip install homegrid
pip install -r requirements.txt --no-deps

ALFWorld

Create conda env.

conda create -n alfworld-ltdt python=3.9
conda activate alfworld-ltdt

Install packages.

cd <path/to/alfworld>
pip install -e .[alfworld]
pip install -r requirements.txt --no-deps

Messenger

Create conda env.

conda create -n messenger python=3.85
conda activate messenger

Install packages

git clone git@github.com:ahjwang/messenger-emma.git
cd <path/to/messenger/messenger-emma>
pip install -e . 
pip install git+https://github.com/ahjwang/py-vgdl
cd <path/to/messenger>
pip install -r requirements.txt --no-deps

Trouble Shoot: If you have difficulty installing packages from messenger-emma for pygame, comment the requirement for pygame version in messenger-emma/setup.py.

MetaWorld

Create conda env.

conda create -n messenger python=3.85
conda activate messenger

cd <path/to/metaworld>
pip install -r requirements.txt --no-deps

🔧 Model Training & Eval

For training and evaluation, please refer to the README in each folder.

• HomeGrid
• ALFWorld
• Messenger
• MetaWorld

🤗 Data & Model ckpts

Data and model checkpoints are available here: Data & Model Checkpoints

Acknowledgement

This work was supported by NSF IIS-1949634 and has benefited from the Microsoft Accelerate Foundation Models Research (AFMR) grant program. We would like to thank the anonymous reviewers for their valuable comments and suggestions.

Citation