Installation

Create Conda Virtual Environment

Recommend to create and activate conda virtual environment with Python 3.7.4

$ conda create --name hetseq
$ conda activate hetseq
$ conda install python=3.7.4

Git Clone and Install Packages

$ git clone https://github.com/yifding/hetseq.git
$ cd /path/to/hetseq
$ pip install -r requirements.txt
$ pip install --editable .

Download BERT Processed File

(Required to run bert model) Download data files including training corpus, model configuration, and BPE dictionary. Test corpus from here, full data from this link. Download test_DATA.zip for test or DATA.zip for full run, unzip it and place the preprocessing/ directory inside the package directory.

Download MNIST Dataset (not required)

(For mnist model) Download MNIST dataset from torchvision, see example here.

from torchvision import datasets
dataset1 = datasets.MNIST('../data', train=True, download=True)