1) There are three folders named compare and train and yml_files. There are two yml files named relcol (use this to run ReLCol) and vcolrl (use this to run VColRL and other baselines)

2) For training, open train folder and run command (with vcolrl environment) : python3 train.py

3) For convenient, we provide the trained model with name 'model_vcolrl.pth' in compare folder. 

4) For inference on VColRL and VColMIS (graphs with less than 10k nodes), open 'compare/table_2_3_ci_vcolmis_vcolrl' and run command (with vcolrl environment) : python3 compare_benchmark.py --mode 1/2

5) For inference on VColRL and VColMIS (graphs with more than 10k nodes), open 'compare/table_2_3_ci_vcolmis_vcolrl_big' and run command (with vcolrl environment) : python3 compare_benchmark.py --mode 1/2

6) For inference on FastColor, open 'compare/table_2_3_ci_fastcolor' and run command (with vcolrl environment) : python3 main_ci.py

7) For inference on ReLCol, open 'compare/table_2_3_relcol' and run command (with relcol environment) : python3 test_learned_policies_on_dataset.py

8) For inference on Gurobi Greedy and TabucolMin, open 'compare/table_2_3_gurobi_greedy_tabucolmin' and run command (with vcolrl environment) : python3 compare_benchmark.py

Note: The DGL library does not support windows platform.




