Skip to content

xuzhiyuan1528/KTM-DRL

main
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Knowledge Transfer in Multi-Task Deep Reinforcement Learning for Continuous Control

This repository is the official implementation of KTM-DRL.

Dependencies

Evaluation

To use the pre-trained models of task-specific teachers and multi-task agent on HalfCheetah task group, download from the Dropbox Link and put the half folder into ./model/ or /your_own_path/.

To evaluate the multi-task agent and its corresponding task-specific teachers, run this command:

python3 eval-ktm.py --seed 0 --cfg ./config/HALF/cfg-mt-half-eval.json --dir ./model/half --name EVA

Training

To train the model(s) in the paper, run this command:

python3 train-ktm.py --seed 0 --cfg ./config/HALF/cfg-mt-half-train.json --dir ./model/half --name TRN

Bibtex

@article{xu2020knowledge,
  title={Knowledge Transfer in Multi-Task Deep Reinforcement Learning for Continuous Control},
  author={Xu, Zhiyuan and Wu, Kun and Che, Zhengping and Tang, Jian and Ye, Jieping},
  journal={arXiv preprint arXiv:2010.07494},
  year={2020}
}

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages