Q&A

• How can one add a new algorithm and benchmark it with GOOD?

Check this issue.

• Can I use my own dataloader?

Yes. In GOOD/data/good_loaders/ folder, create a new python file and inherit the basic loader class in BaseLoader.py. Don’t forget to register it.

• Can I use more flexible pipeline structure?

Yes. In GOOD/kernel/pipelines/ folder, create a new python file or copy the basic_pipeline.py, then build or modify your own pipeline! It is much easier to copy and modify the original pipeline than building a new one from scratch unless you are completely familiar with this project.

For more questions, please submit issues or open discussions in our GitHub repo.