This project provides the official implementation of the CLOUD agent introduced in "CLOUD: Contrastive Learning of Unsupervised Dynamics", by Jianren Wang*, Yujie Lu* and Hang Zhao. (* indicates equal contribution)
CLOUD Architecture: Our framework consists of four learnable functions, including a forward dynamics model F(·), an inverse dynamics model I(·), a state representation model g(·) and an action representation model q(·). We propose to learn these four models jointly via contrastive estimation.
Official datasets, samplers and processors(to be released). You can also use the 'Poke' dataset available in here.
Please refer to the following script to train the default cloud agent.
python train_cloud.py -batch_size 256 -lrt 0.001 -save_dir /path/to/save