based on https://github.com/ALRhub/cw2

I found one problem. I tried to experiment with the arxiv_2023 datasets for different Samplers, but my process was stopped. The problem is that the train dataset is very large. The problem is with train/opt_train.py in this function, but maybe I have these problems because I'm doing my experiments on the CPU

utility of gpu is lower than 3%

write the mathematical formulation of the link prediction with optimization problem

list out the loss function with the data distribution

1. create dataset download script

2. trainer and documentation for generated embedding with documentation

1. visualization of the original graph for Heterogeneity
   1.1 cumulative degree distribution
   1.2 predictability analysis

2. Performance vs Metrics correlation plot between hits@k and auc