Implementations of Tensor networks for unsupervised machine learning in PyTorch.
The code is tested on RTX 2080Ti and Tesla V100 with Python 3.6.8, PyTorch 1.4.0 and CUDA 10.1.
The src_amps contains code for AMPS, the src_conv_amps contains code for Convolution AMPS, the src_rbm contains code for RMB and the src_rwd contains code for real world datasets experiments. To run the MNIST example, please download binarized mnist dataset and put it in ./data.
The implementations of MPS and LPS can be found in tensor_networks_for_probabilistic_modeling. The implementations of VAN can be found in stat-mech-van.